{
  "llms": [
    {
      "id": "~anthropic/claude-fable-latest",
      "orId": "~anthropic/claude-fable-latest",
      "vendor": "~anthropic",
      "model": "claude-fable-latest",
      "name": "Anthropic: Claude Fable Latest",
      "blurb": "This model always redirects to the latest model in the Claude Fable family.",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 16,
        "output": 80,
        "cacheRead": 0.4,
        "cacheWrite": 20,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-06-09"
    },
    {
      "id": "~anthropic/claude-haiku-latest",
      "orId": "~anthropic/claude-haiku-latest",
      "vendor": "~anthropic",
      "model": "claude-haiku-latest",
      "name": "Anthropic: Claude Haiku Latest",
      "blurb": "This model always redirects to the latest model in the Claude Haiku family.",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.2,
        "output": 1,
        "cacheRead": 0.02,
        "cacheWrite": 0.25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 100000,
            "input": 1,
            "output": 5,
            "cacheRead": 0.1
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-04-27"
    },
    {
      "id": "~anthropic/claude-opus-latest",
      "orId": "~anthropic/claude-opus-latest",
      "vendor": "~anthropic",
      "model": "claude-opus-latest",
      "name": "Anthropic: Claude Opus Latest",
      "blurb": "This model always redirects to the latest model in the Claude Opus family.",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 8,
        "output": 40,
        "cacheRead": 0.4,
        "cacheWrite": 10,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-04-21"
    },
    {
      "id": "~anthropic/claude-sonnet-latest",
      "orId": "~anthropic/claude-sonnet-latest",
      "vendor": "~anthropic",
      "model": "claude-sonnet-latest",
      "name": "Anthropic: Claude Sonnet Latest",
      "blurb": "This model always redirects to the latest model in the Claude Sonnet family.",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.2,
        "cacheWrite": 5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-04-27"
    },
    {
      "id": "~deepseek/deepseek-flash-latest",
      "orId": "~deepseek/deepseek-flash-latest",
      "vendor": "~deepseek",
      "model": "deepseek-flash-latest",
      "name": "DeepSeek: DeepSeek Flash Latest",
      "blurb": "This model always redirects to the latest model in the DeepSeek Flash family.",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.072,
        "output": 1,
        "cacheRead": 0.02
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-14"
    },
    {
      "id": "~deepseek/deepseek-pro-latest",
      "orId": "~deepseek/deepseek-pro-latest",
      "vendor": "~deepseek",
      "model": "deepseek-pro-latest",
      "name": "DeepSeek: DeepSeek Pro Latest",
      "blurb": "This model always redirects to the latest model in the DeepSeek Pro family.",
      "contextWindow": 1048576,
      "maxOutput": 393216,
      "cost": {
        "input": 0.1998,
        "output": 16,
        "cacheRead": 0.1998
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-14"
    },
    {
      "id": "~deepseek/deepseek-v4-flash-latest",
      "orId": "~deepseek/deepseek-v4-flash-latest",
      "vendor": "~deepseek",
      "model": "deepseek-v4-flash-latest",
      "name": "DeepSeek: DeepSeek V4 Flash Latest",
      "blurb": "This model always redirects to the latest model in the DeepSeek V4 Flash family.",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.0868,
        "output": 0.6,
        "cacheRead": 0.07
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-08-01"
    },
    {
      "id": "~google/gemini-flash-latest",
      "orId": "~google/gemini-flash-latest",
      "vendor": "~google",
      "model": "gemini-flash-latest",
      "name": "Google: Gemini Flash Latest",
      "blurb": "This model always redirects to the latest model in the Gemini Flash family.",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 3,
        "output": 15,
        "cacheRead": 0.15,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 3
      },
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-04-27"
    },
    {
      "id": "~google/gemini-pro-latest",
      "orId": "~google/gemini-pro-latest",
      "vendor": "~google",
      "model": "gemini-pro-latest",
      "name": "Google: Gemini Pro Latest",
      "blurb": "This model always redirects to the latest model in the Gemini Pro family.",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.2,
        "cacheWrite": 0.375,
        "webSearch": 0.028,
        "audioIn": 4
      },
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-04-27"
    },
    {
      "id": "~moonshotai/kimi-latest",
      "orId": "~moonshotai/kimi-latest",
      "vendor": "~moonshotai",
      "model": "kimi-latest",
      "name": "MoonshotAI: Kimi Latest",
      "blurb": "This model always redirects to the latest model in the Kimi family.",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.3648,
        "output": 22.011,
        "cacheRead": 0.36
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-04-27"
    },
    {
      "id": "~openai/gpt-astra-latest",
      "orId": "~openai/gpt-astra-latest",
      "vendor": "~openai",
      "model": "gpt-astra-latest",
      "name": "OpenAI: GPT Astra Latest",
      "blurb": "This model always redirects to the latest model in the GPT Astra family.",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 20,
        "output": 100,
        "cacheRead": 2,
        "cacheWrite": 25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 40,
            "output": 150,
            "cacheRead": 4
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-11"
    },
    {
      "id": "~openai/gpt-luna-latest",
      "orId": "~openai/gpt-luna-latest",
      "vendor": "~openai",
      "model": "gpt-luna-latest",
      "name": "OpenAI: GPT Luna Latest",
      "blurb": "This model always redirects to the latest model in the GPT Luna family.",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.2,
        "output": 1,
        "cacheRead": 0.02,
        "cacheWrite": 0.25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 0.4,
            "output": 1.5,
            "cacheRead": 0.04
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-11"
    },
    {
      "id": "~openai/gpt-mini-latest",
      "orId": "~openai/gpt-mini-latest",
      "vendor": "~openai",
      "model": "gpt-mini-latest",
      "name": "OpenAI: GPT Mini Latest",
      "blurb": "This model always redirects to the latest model in the GPT Mini family.",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 1.5,
        "output": 9,
        "cacheRead": 0.15,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-08-31",
      "released": "2026-04-27"
    },
    {
      "id": "~openai/gpt-sol-latest",
      "orId": "~openai/gpt-sol-latest",
      "vendor": "~openai",
      "model": "gpt-sol-latest",
      "name": "OpenAI: GPT Sol Latest",
      "blurb": "This model always redirects to the latest model in the GPT Sol family.",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.2,
        "cacheWrite": 5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 8,
            "output": 30,
            "cacheRead": 0.4
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-11"
    },
    {
      "id": "~openai/gpt-terra-latest",
      "orId": "~openai/gpt-terra-latest",
      "vendor": "~openai",
      "model": "gpt-terra-latest",
      "name": "OpenAI: GPT Terra Latest",
      "blurb": "This model always redirects to the latest model in the GPT Terra family.",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.1,
        "cacheWrite": 2.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 4,
            "output": 15,
            "cacheRead": 0.2
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-11"
    },
    {
      "id": "~unbiased/pareto-latest",
      "orId": "~unbiased/pareto-latest",
      "vendor": "~unbiased",
      "model": "pareto-latest",
      "name": "Pareto Latest",
      "blurb": "This model always redirects to the latest model in the Pareto family.",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 1.6,
        "output": 6.4,
        "cacheRead": 0.06
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2026-10-13"
    },
    {
      "id": "~x-ai/grok-latest",
      "orId": "~x-ai/grok-latest",
      "vendor": "~x-ai",
      "model": "grok-latest",
      "name": "xAI: Grok Latest",
      "blurb": "This model always redirects to the latest Grok model from xAI.",
      "contextWindow": 500000,
      "maxOutput": 450000,
      "cost": {
        "input": 4,
        "output": 12,
        "cacheRead": 1,
        "cacheWrite": 0,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 8,
            "output": 24,
            "cacheRead": 2
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-07-08"
    },
    {
      "id": "~z-ai/glm-flash-latest",
      "orId": "~z-ai/glm-flash-latest",
      "vendor": "~z-ai",
      "model": "glm-flash-latest",
      "name": "Z.ai: GLM Flash Latest",
      "blurb": "This model always redirects to the latest model in the GLM Flash family.",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0.0798,
        "output": 0.1568,
        "cacheRead": 0.06
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-08-27"
    },
    {
      "id": "~z-ai/glm-latest",
      "orId": "~z-ai/glm-latest",
      "vendor": "~z-ai",
      "model": "glm-latest",
      "name": "Z.ai: GLM Latest",
      "blurb": "This model always redirects to the latest GLM model from Z.ai.",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.1398,
        "output": 2.8,
        "cacheRead": 0.1378
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-08-19"
    },
    {
      "id": "aion-labs/aion-2.0",
      "orId": "aion-labs/aion-2.0",
      "vendor": "aion-labs",
      "model": "aion-2.0",
      "name": "AionLabs: Aion-2.0",
      "blurb": "Aion-2.0 is a variant of DeepSeek V3.2 optimized for immersive roleplaying and storytelling. It is particularly strong at introducing tension, crises, and conflict into stories, making narratives feel more engaging....",
      "contextWindow": 131072,
      "maxOutput": 32768,
      "cost": {
        "input": 1.6,
        "output": 3.2,
        "cacheRead": 0.4
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-02-23"
    },
    {
      "id": "aion-labs/aion-3.0",
      "orId": "aion-labs/aion-3.0",
      "vendor": "aion-labs",
      "model": "aion-3.0",
      "name": "AionLabs: Aion-3.0",
      "blurb": "Aion-3.0 is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models. It uses a collaborative generation process in which multiple specialized models each contribute...",
      "contextWindow": 131072,
      "maxOutput": 32768,
      "cost": {
        "input": 6,
        "output": 12,
        "cacheRead": 1.5
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-07-07"
    },
    {
      "id": "aion-labs/aion-3.0-mini",
      "orId": "aion-labs/aion-3.0-mini",
      "vendor": "aion-labs",
      "model": "aion-3.0-mini",
      "name": "AionLabs: Aion-3.0-Mini",
      "blurb": "Aion-3.0 Mini is a multi-model roleplaying and storytelling system from AionLabs, built on the DeepSeek family of models. It uses a collaborative generation process in which multiple specialized models each...",
      "contextWindow": 131072,
      "maxOutput": 32768,
      "cost": {
        "input": 1.4,
        "output": 2.8,
        "cacheRead": 0.36
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-07-07"
    },
    {
      "id": "aion-labs/aion-3.5",
      "orId": "aion-labs/aion-3.5",
      "vendor": "aion-labs",
      "model": "aion-3.5",
      "name": "AionLabs: Aion 3.5",
      "blurb": "Aion 3.5 is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models. It uses a collaborative generation process in which multiple specialized models each...",
      "contextWindow": 262144,
      "maxOutput": 32768,
      "cost": {
        "input": 6,
        "output": 12,
        "cacheRead": 1.5
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-23"
    },
    {
      "id": "aion-labs/aion-3.5-mini",
      "orId": "aion-labs/aion-3.5-mini",
      "vendor": "aion-labs",
      "model": "aion-3.5-mini",
      "name": "AionLabs: Aion 3.5 Mini",
      "blurb": "Aion 3.5 Mini is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models. It is the smaller, lower-cost sibling of Aion 3.5 and uses...",
      "contextWindow": 262144,
      "maxOutput": 32768,
      "cost": {
        "input": 1.4,
        "output": 2.8,
        "cacheRead": 0.36
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-23"
    },
    {
      "id": "aion-labs/aion-rp-llama-3.1-8b",
      "orId": "aion-labs/aion-rp-llama-3.1-8b",
      "vendor": "aion-labs",
      "model": "aion-rp-llama-3.1-8b",
      "name": "AionLabs: Aion-RP 1.0 (8B)",
      "blurb": "Aion-RP-Llama-3.1-8B ranks the highest in the character evaluation portion of the RPBench-Auto benchmark, a roleplaying-specific variant of Arena-Hard-Auto, where LLMs evaluate each other’s responses. It is a fine-tuned base model...",
      "contextWindow": 32768,
      "maxOutput": 29491,
      "cost": {
        "input": 1.6,
        "output": 3.2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2023-12-31",
      "released": "2025-02-04",
      "openWeights": {
        "huggingFace": "aion-labs/Aion-RP-Llama-3.1-8B"
      }
    },
    {
      "id": "amazon/nova-2-lite-v1",
      "orId": "amazon/nova-2-lite-v1",
      "vendor": "amazon",
      "model": "nova-2-lite-v1",
      "name": "Amazon: Nova 2 Lite",
      "blurb": "Nova 2 Lite is a fast, cost-effective reasoning model for everyday workloads that can process text, images, and videos to generate text. Nova 2 Lite demonstrates standout capabilities in processing...",
      "contextWindow": 1000000,
      "maxOutput": 65535,
      "cost": {
        "input": 0.6,
        "output": 5
      },
      "input": [
        "text",
        "image",
        "video",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2025-12-02",
      "regions": [
        "eu"
      ]
    },
    {
      "id": "amazon/nova-lite-v1",
      "orId": "amazon/nova-lite-v1",
      "vendor": "amazon",
      "model": "nova-lite-v1",
      "name": "Amazon: Nova Lite 1.0",
      "blurb": "Amazon Nova Lite 1.0 is a very low-cost multimodal model from Amazon that focused on fast processing of image, video, and text inputs to generate text output. Amazon Nova Lite...",
      "contextWindow": 300000,
      "maxOutput": 5120,
      "cost": {
        "input": 0.12,
        "output": 0.48
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-10-31",
      "released": "2024-12-05",
      "regions": [
        "eu"
      ]
    },
    {
      "id": "amazon/nova-micro-v1",
      "orId": "amazon/nova-micro-v1",
      "vendor": "amazon",
      "model": "nova-micro-v1",
      "name": "Amazon: Nova Micro 1.0",
      "blurb": "Amazon Nova Micro 1.0 is a text-only model that delivers the lowest latency responses in the Amazon Nova family of models at a very low cost. With a context length...",
      "contextWindow": 128000,
      "maxOutput": 5120,
      "cost": {
        "input": 0.07,
        "output": 0.28
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-10-31",
      "released": "2024-12-05",
      "regions": [
        "eu"
      ]
    },
    {
      "id": "amazon/nova-premier-v1",
      "orId": "amazon/nova-premier-v1",
      "vendor": "amazon",
      "model": "nova-premier-v1",
      "name": "Amazon: Nova Premier 1.0",
      "blurb": "Amazon Nova Premier is the most capable of Amazon’s multimodal models for complex reasoning tasks and for use as the best teacher for distilling custom models.",
      "contextWindow": 1000000,
      "maxOutput": 32000,
      "cost": {
        "input": 5,
        "output": 25,
        "cacheRead": 1.25
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2025-10-31"
    },
    {
      "id": "amazon/nova-pro-v1",
      "orId": "amazon/nova-pro-v1",
      "vendor": "amazon",
      "model": "nova-pro-v1",
      "name": "Amazon: Nova Pro 1.0",
      "blurb": "Amazon Nova Pro 1.0 is a capable multimodal model from Amazon focused on providing a combination of accuracy, speed, and cost for a wide range of tasks. As of December...",
      "contextWindow": 300000,
      "maxOutput": 5120,
      "cost": {
        "input": 1.6,
        "output": 6.4
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-10-31",
      "released": "2024-12-05",
      "regions": [
        "eu"
      ]
    },
    {
      "id": "anthracite-org/magnum-v4-72b",
      "orId": "anthracite-org/magnum-v4-72b",
      "vendor": "anthracite-org",
      "model": "magnum-v4-72b",
      "name": "Magnum v4 72B",
      "blurb": "This is a series of models designed to replicate the prose quality of the Claude 3 models, specifically Sonnet() and Opus().\n\nThe model is fine-tuned on top of Qwen2.5 72B.",
      "contextWindow": 32768,
      "maxOutput": 4096,
      "cost": {
        "input": 5,
        "output": 10,
        "cacheRead": 5
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-06-30",
      "released": "2024-10-22",
      "openWeights": {
        "huggingFace": "anthracite-org/magnum-v4-72b"
      }
    },
    {
      "id": "anthropic/claude-fable-5",
      "orId": "anthropic/claude-fable-5",
      "vendor": "anthropic",
      "model": "claude-fable-5",
      "name": "Anthropic: Claude Fable 5",
      "blurb": "Claude Fable 5 is a Mythos-class model from Anthropic, built for autonomous knowledge work and coding. It supports text, image, and file inputs with text output, with reasoning support and...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 20,
        "output": 100,
        "cacheRead": 2,
        "cacheWrite": 25,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-06-09",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 164.13,
        "accessibility": "API access",
        "gpqa": 0.859,
        "gpqaStderr": 0.0248,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.707,
        "simpleQAStderr": 0.0144
      }
    },
    {
      "id": "anthropic/claude-fable-5:batch",
      "orId": "anthropic/claude-fable-5:batch",
      "vendor": "anthropic",
      "model": "claude-fable-5:batch",
      "name": "Anthropic: Claude Fable 5 (batch)",
      "blurb": "Claude Fable 5 is a Mythos-class model from Anthropic, built for autonomous knowledge work and coding. It supports text, image, and file inputs with text output, with reasoning support and...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 50,
        "cacheRead": 1,
        "cacheWrite": 12.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-06-09"
    },
    {
      "id": "anthropic/claude-fable-5.1",
      "orId": "anthropic/claude-fable-5.1",
      "vendor": "anthropic",
      "model": "claude-fable-5.1",
      "name": "Anthropic: Claude Fable 5.1",
      "blurb": "Claude Fable 5.1 improves on Claude Fable 5 across the board, with the biggest gains in agentic coding, long-running agentic workflows, and knowledge work: long code refactors, front-end and visual...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 20,
        "output": 100,
        "cacheRead": 0.5,
        "cacheWrite": 25,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-01",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 166.99,
        "accessibility": "API access",
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.708,
        "simpleQAStderr": 0.0144
      }
    },
    {
      "id": "anthropic/claude-fable-5.1:batch",
      "orId": "anthropic/claude-fable-5.1:batch",
      "vendor": "anthropic",
      "model": "claude-fable-5.1:batch",
      "name": "Anthropic: Claude Fable 5.1 (batch)",
      "blurb": "Claude Fable 5.1 improves on Claude Fable 5 across the board, with the biggest gains in agentic coding, long-running agentic workflows, and knowledge work: long code refactors, front-end and visual...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 50,
        "cacheRead": 0.25,
        "cacheWrite": 12.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-01"
    },
    {
      "id": "anthropic/claude-fable-5.2",
      "orId": "anthropic/claude-fable-5.2",
      "vendor": "anthropic",
      "model": "claude-fable-5.2",
      "name": "Anthropic: Claude Fable 5.2",
      "blurb": "Claude Fable 5.2 is Anthropic's most capable model for the most demanding technical work, with gains over Fable 5.1 in coding, reasoning, and vision. In prerelease testing it completed otherwise-infeasible...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 16,
        "output": 80,
        "cacheRead": 0.4,
        "cacheWrite": 20,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-11-09",
      "bench": {
        "source": "epoch",
        "eci": 173.44,
        "accessibility": "API access",
        "gpqa": 0.939,
        "gpqaStderr": 0.0163,
        "aime": 0.997,
        "aimeStderr": 0.0028,
        "simpleQA": 0.94,
        "simpleQAStderr": 0.0075
      }
    },
    {
      "id": "anthropic/claude-fable-5.2:batch",
      "orId": "anthropic/claude-fable-5.2:batch",
      "vendor": "anthropic",
      "model": "claude-fable-5.2:batch",
      "name": "Anthropic: Claude Fable 5.2 (batch)",
      "blurb": "Claude Fable 5.2 is Anthropic's most capable model for the most demanding technical work, with gains over Fable 5.1 in coding, reasoning, and vision. In prerelease testing it completed otherwise-infeasible...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 8,
        "output": 40,
        "cacheRead": 0.2,
        "cacheWrite": 10,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-11-09"
    },
    {
      "id": "anthropic/claude-haiku-4.5",
      "orId": "anthropic/claude-haiku-4.5",
      "vendor": "anthropic",
      "model": "claude-haiku-4.5",
      "name": "Anthropic: Claude Haiku 4.5",
      "blurb": "Claude Haiku 4.5 is Anthropic’s fastest and most efficient model, delivering near-frontier intelligence at a fraction of the cost and latency of larger Claude models. Matching Claude Sonnet 4’s performance...",
      "contextWindow": 200000,
      "maxOutput": 64000,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.2,
        "cacheWrite": 2.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2025-10-15",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 142.83,
        "accessibility": "API access"
      }
    },
    {
      "id": "anthropic/claude-haiku-4.5:batch",
      "orId": "anthropic/claude-haiku-4.5:batch",
      "vendor": "anthropic",
      "model": "claude-haiku-4.5:batch",
      "name": "Anthropic: Claude Haiku 4.5 (batch)",
      "blurb": "Claude Haiku 4.5 is Anthropic’s fastest and most efficient model, delivering near-frontier intelligence at a fraction of the cost and latency of larger Claude models. Matching Claude Sonnet 4’s performance...",
      "contextWindow": 200000,
      "maxOutput": 64000,
      "cost": {
        "input": 1,
        "output": 5,
        "cacheRead": 0.1,
        "cacheWrite": 1.25,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2025-10-15"
    },
    {
      "id": "anthropic/claude-haiku-5.5",
      "orId": "anthropic/claude-haiku-5.5",
      "vendor": "anthropic",
      "model": "claude-haiku-5.5",
      "name": "Anthropic: Claude Haiku 5.5",
      "blurb": "Claude Haiku 5.5 is Anthropic's small, fast model for high-volume, cost-sensitive work such as summarization, subagents, and browser use. It succeeds Claude Haiku 4.5 with stronger coding, computer use, and...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.2,
        "output": 1,
        "cacheRead": 0.02,
        "cacheWrite": 0.25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 100000,
            "input": 1,
            "output": 5,
            "cacheRead": 0.1
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-07",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 160.43,
        "accessibility": "API access",
        "gpqa": 0.896,
        "gpqaStderr": 0.0198,
        "aime": 0.989,
        "aimeStderr": 0.0054,
        "simpleQA": 0.238,
        "simpleQAStderr": 0.0135
      }
    },
    {
      "id": "anthropic/claude-haiku-5.5:batch",
      "orId": "anthropic/claude-haiku-5.5:batch",
      "vendor": "anthropic",
      "model": "claude-haiku-5.5:batch",
      "name": "Anthropic: Claude Haiku 5.5 (batch)",
      "blurb": "Claude Haiku 5.5 is Anthropic's small, fast model for high-volume, cost-sensitive work such as summarization, subagents, and browser use. It succeeds Claude Haiku 4.5 with stronger coding, computer use, and...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.1,
        "output": 0.5,
        "cacheRead": 0.01,
        "cacheWrite": 0.125,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 100000,
            "input": 0.5,
            "output": 2.5,
            "cacheRead": 0.05
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-07"
    },
    {
      "id": "anthropic/claude-opus-4.1",
      "orId": "anthropic/claude-opus-4.1",
      "vendor": "anthropic",
      "model": "claude-opus-4.1",
      "name": "Anthropic: Claude Opus 4.1",
      "blurb": "Claude Opus 4.1 is an updated version of Anthropic’s flagship model, offering improved performance in coding, reasoning, and agentic tasks. It achieves 74.5% on SWE-bench Verified and shows notable gains...",
      "contextWindow": 200000,
      "maxOutput": 32000,
      "cost": {
        "input": 30,
        "output": 150,
        "cacheRead": 3,
        "cacheWrite": 37.5,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-08-05",
      "bench": {
        "source": "epoch",
        "eci": 144.84,
        "accessibility": "API access"
      }
    },
    {
      "id": "anthropic/claude-opus-4.1:batch",
      "orId": "anthropic/claude-opus-4.1:batch",
      "vendor": "anthropic",
      "model": "claude-opus-4.1:batch",
      "name": "Anthropic: Claude Opus 4.1 (batch)",
      "blurb": "Claude Opus 4.1 is an updated version of Anthropic’s flagship model, offering improved performance in coding, reasoning, and agentic tasks. It achieves 74.5% on SWE-bench Verified and shows notable gains...",
      "contextWindow": 200000,
      "maxOutput": 32000,
      "cost": {
        "input": 15,
        "output": 75,
        "cacheRead": 1.5,
        "cacheWrite": 18.75,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-08-05"
    },
    {
      "id": "anthropic/claude-opus-4.5",
      "orId": "anthropic/claude-opus-4.5",
      "vendor": "anthropic",
      "model": "claude-opus-4.5",
      "name": "Anthropic: Claude Opus 4.5",
      "blurb": "Claude Opus 4.5 is Anthropic’s frontier reasoning model optimized for complex software engineering, agentic workflows, and long-horizon computer use. It offers strong multimodal capabilities, competitive performance across real-world coding and...",
      "contextWindow": 200000,
      "maxOutput": 64000,
      "cost": {
        "input": 10,
        "output": 50,
        "cacheRead": 1,
        "cacheWrite": 12.5,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2025-11-24",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 151.92,
        "accessibility": "API access"
      }
    },
    {
      "id": "anthropic/claude-opus-4.5:batch",
      "orId": "anthropic/claude-opus-4.5:batch",
      "vendor": "anthropic",
      "model": "claude-opus-4.5:batch",
      "name": "Anthropic: Claude Opus 4.5 (batch)",
      "blurb": "Claude Opus 4.5 is Anthropic’s frontier reasoning model optimized for complex software engineering, agentic workflows, and long-horizon computer use. It offers strong multimodal capabilities, competitive performance across real-world coding and...",
      "contextWindow": 200000,
      "maxOutput": 64000,
      "cost": {
        "input": 5,
        "output": 25,
        "cacheRead": 0.5,
        "cacheWrite": 6.25,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2025-11-24"
    },
    {
      "id": "anthropic/claude-opus-4.6",
      "orId": "anthropic/claude-opus-4.6",
      "vendor": "anthropic",
      "model": "claude-opus-4.6",
      "name": "Anthropic: Claude Opus 4.6",
      "blurb": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire workflows rather than single prompts, making it especially effective...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 50,
        "cacheRead": 1,
        "cacheWrite": 12.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-02-04",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 156.34,
        "accessibility": "API access",
        "gpqa": 0.905,
        "gpqaStderr": 0.0169,
        "sweBenchVerified": 0.787,
        "sweBenchVerifiedStderr": 0.0186,
        "frontierMath": 0.407,
        "frontierMathStderr": 0.029,
        "aime": 0.944,
        "aimeStderr": 0.0282,
        "simpleQA": 0.47,
        "simpleQAStderr": 0.0158
      }
    },
    {
      "id": "anthropic/claude-opus-4.6:batch",
      "orId": "anthropic/claude-opus-4.6:batch",
      "vendor": "anthropic",
      "model": "claude-opus-4.6:batch",
      "name": "Anthropic: Claude Opus 4.6 (batch)",
      "blurb": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire workflows rather than single prompts, making it especially effective...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 5,
        "output": 25,
        "cacheRead": 0.5,
        "cacheWrite": 6.25,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-02-04"
    },
    {
      "id": "anthropic/claude-opus-4.7",
      "orId": "anthropic/claude-opus-4.7",
      "vendor": "anthropic",
      "model": "claude-opus-4.7",
      "name": "Anthropic: Claude Opus 4.7",
      "blurb": "Opus 4.7 is the next generation of Anthropic's Opus family, built for long-running, asynchronous agents. Building on the coding and agentic strengths of Opus 4.6, it delivers stronger performance on...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 50,
        "cacheRead": 1,
        "cacheWrite": 12.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-04-16",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 157.44,
        "accessibility": "API access",
        "gpqa": 0.902,
        "gpqaStderr": 0.0185,
        "sweBenchVerified": 0.835,
        "sweBenchVerifiedStderr": 0.0169,
        "frontierMath": 0.438,
        "frontierMathStderr": 0.0292,
        "aime": 0.978,
        "aimeStderr": 0.022,
        "simpleQA": 0.517,
        "simpleQAStderr": 0.0158
      }
    },
    {
      "id": "anthropic/claude-opus-4.7:batch",
      "orId": "anthropic/claude-opus-4.7:batch",
      "vendor": "anthropic",
      "model": "claude-opus-4.7:batch",
      "name": "Anthropic: Claude Opus 4.7 (batch)",
      "blurb": "Opus 4.7 is the next generation of Anthropic's Opus family, built for long-running, asynchronous agents. Building on the coding and agentic strengths of Opus 4.6, it delivers stronger performance on...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 5,
        "output": 25,
        "cacheRead": 0.5,
        "cacheWrite": 6.25,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-04-16"
    },
    {
      "id": "anthropic/claude-opus-4.8",
      "orId": "anthropic/claude-opus-4.8",
      "vendor": "anthropic",
      "model": "claude-opus-4.8",
      "name": "Anthropic: Claude Opus 4.8",
      "blurb": "Claude Opus 4.8 is Anthropic's most capable generally available model in the Opus family. It supports text, image, and file inputs with text output, with reasoning support and a 1M-token...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 50,
        "cacheRead": 1,
        "cacheWrite": 12.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-05-27",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 159.94,
        "accessibility": "API access",
        "gpqa": 0.91,
        "gpqaStderr": 0.0192,
        "frontierMath": 0.472,
        "frontierMathStderr": 0.0294,
        "aime": 0.983,
        "aimeStderr": 0.0141,
        "simpleQA": 0.53,
        "simpleQAStderr": 0.0158
      }
    },
    {
      "id": "anthropic/claude-opus-4.8:batch",
      "orId": "anthropic/claude-opus-4.8:batch",
      "vendor": "anthropic",
      "model": "claude-opus-4.8:batch",
      "name": "Anthropic: Claude Opus 4.8 (batch)",
      "blurb": "Claude Opus 4.8 is Anthropic's most capable generally available model in the Opus family. It supports text, image, and file inputs with text output, with reasoning support and a 1M-token...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 5,
        "output": 25,
        "cacheRead": 0.5,
        "cacheWrite": 6.25,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-05-27"
    },
    {
      "id": "anthropic/claude-opus-5",
      "orId": "anthropic/claude-opus-5",
      "vendor": "anthropic",
      "model": "claude-opus-5",
      "name": "Anthropic: Claude Opus 5",
      "blurb": "Claude Opus 5 is Anthropic’s flagship model for demanding reasoning, coding, and long-horizon agentic work. It is particularly strong at end-to-end software tasks, code review and bug finding, visual analysis...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 50,
        "cacheRead": 1,
        "cacheWrite": 12.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-07-24",
      "recommended": true,
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 164.68,
        "accessibility": "API access",
        "gpqa": 0.939,
        "gpqaStderr": 0.0148,
        "aime": 0.989,
        "aimeStderr": 0.0111,
        "simpleQA": 0.599,
        "simpleQAStderr": 0.0155
      }
    },
    {
      "id": "anthropic/claude-opus-5:batch",
      "orId": "anthropic/claude-opus-5:batch",
      "vendor": "anthropic",
      "model": "claude-opus-5:batch",
      "name": "Anthropic: Claude Opus 5 (batch)",
      "blurb": "Claude Opus 5 is Anthropic’s flagship model for demanding reasoning, coding, and long-horizon agentic work. It is particularly strong at end-to-end software tasks, code review and bug finding, visual analysis...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 5,
        "output": 25,
        "cacheRead": 0.5,
        "cacheWrite": 6.25,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-07-24"
    },
    {
      "id": "anthropic/claude-opus-5.5",
      "orId": "anthropic/claude-opus-5.5",
      "vendor": "anthropic",
      "model": "claude-opus-5.5",
      "name": "Anthropic: Claude Opus 5.5",
      "blurb": "Claude Opus 5.5 is Anthropic's flagship model for demanding reasoning, coding, and long-horizon agentic work, succeeding Claude Opus 5. It is particularly strong at multi-step changes in large codebases, code...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 8,
        "output": 40,
        "cacheRead": 0.4,
        "cacheWrite": 10,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-22",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 168.54,
        "accessibility": "API access",
        "gpqa": 0.906,
        "gpqaStderr": 0.0194,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.722,
        "simpleQAStderr": 0.0142
      }
    },
    {
      "id": "anthropic/claude-opus-5.5:batch",
      "orId": "anthropic/claude-opus-5.5:batch",
      "vendor": "anthropic",
      "model": "claude-opus-5.5:batch",
      "name": "Anthropic: Claude Opus 5.5 (batch)",
      "blurb": "Claude Opus 5.5 is Anthropic's flagship model for demanding reasoning, coding, and long-horizon agentic work, succeeding Claude Opus 5. It is particularly strong at multi-step changes in large codebases, code...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.2,
        "cacheWrite": 5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-22"
    },
    {
      "id": "anthropic/claude-sonnet-4.5",
      "orId": "anthropic/claude-sonnet-4.5",
      "vendor": "anthropic",
      "model": "claude-sonnet-4.5",
      "name": "Anthropic: Claude Sonnet 4.5",
      "blurb": "Claude Sonnet 4.5 is Anthropic’s most advanced Sonnet model to date, optimized for real-world agents and coding workflows. It delivers state-of-the-art performance on coding benchmarks such as SWE-bench Verified, with...",
      "contextWindow": 1000000,
      "maxOutput": 64000,
      "cost": {
        "input": 6,
        "output": 30,
        "cacheRead": 0.6,
        "cacheWrite": 7.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 12,
            "output": 45,
            "cacheRead": 1.2
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-09-29",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 147.63,
        "accessibility": "API access"
      }
    },
    {
      "id": "anthropic/claude-sonnet-4.5:batch",
      "orId": "anthropic/claude-sonnet-4.5:batch",
      "vendor": "anthropic",
      "model": "claude-sonnet-4.5:batch",
      "name": "Anthropic: Claude Sonnet 4.5 (batch)",
      "blurb": "Claude Sonnet 4.5 is Anthropic’s most advanced Sonnet model to date, optimized for real-world agents and coding workflows. It delivers state-of-the-art performance on coding benchmarks such as SWE-bench Verified, with...",
      "contextWindow": 1000000,
      "maxOutput": 64000,
      "cost": {
        "input": 3,
        "output": 15,
        "cacheRead": 0.3,
        "cacheWrite": 3.75,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 6,
            "output": 22.5,
            "cacheRead": 0.6
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-09-29"
    },
    {
      "id": "anthropic/claude-sonnet-4.6",
      "orId": "anthropic/claude-sonnet-4.6",
      "vendor": "anthropic",
      "model": "claude-sonnet-4.6",
      "name": "Anthropic: Claude Sonnet 4.6",
      "blurb": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It excels at iterative development, complex codebase navigation, end-to-end project management with...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 6,
        "output": 30,
        "cacheRead": 0.6,
        "cacheWrite": 7.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-02-17",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 152.86,
        "accessibility": "API access",
        "gpqa": 0.874,
        "gpqaStderr": 0.02,
        "sweBenchVerified": 0.752,
        "sweBenchVerifiedStderr": 0.0196,
        "frontierMath": 0.324,
        "frontierMathStderr": 0.0281,
        "aime": 0.858,
        "aimeStderr": 0.0454,
        "simpleQA": 0.355,
        "simpleQAStderr": 0.0151
      }
    },
    {
      "id": "anthropic/claude-sonnet-4.6:batch",
      "orId": "anthropic/claude-sonnet-4.6:batch",
      "vendor": "anthropic",
      "model": "claude-sonnet-4.6:batch",
      "name": "Anthropic: Claude Sonnet 4.6 (batch)",
      "blurb": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It excels at iterative development, complex codebase navigation, end-to-end project management with...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 3,
        "output": 15,
        "cacheRead": 0.3,
        "cacheWrite": 3.75,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-02-17"
    },
    {
      "id": "anthropic/claude-sonnet-5",
      "orId": "anthropic/claude-sonnet-5",
      "vendor": "anthropic",
      "model": "claude-sonnet-5",
      "name": "Anthropic: Claude Sonnet 5",
      "blurb": "Sonnet 5 is Anthropic's most capable Sonnet-class model, with frontier performance across coding, agents, and professional work. It supports adaptive thinking with selectable reasoning effort levels (low, medium, high, max,...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.4,
        "cacheWrite": 5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-06-30",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 156.94,
        "accessibility": "API access",
        "gpqa": 0.905,
        "gpqaStderr": 0.0178,
        "aime": 0.947,
        "aimeStderr": 0.0226,
        "simpleQA": 0.337,
        "simpleQAStderr": 0.015
      }
    },
    {
      "id": "anthropic/claude-sonnet-5:batch",
      "orId": "anthropic/claude-sonnet-5:batch",
      "vendor": "anthropic",
      "model": "claude-sonnet-5:batch",
      "name": "Anthropic: Claude Sonnet 5 (batch)",
      "blurb": "Sonnet 5 is Anthropic's most capable Sonnet-class model, with frontier performance across coding, agents, and professional work. It supports adaptive thinking with selectable reasoning effort levels (low, medium, high, max,...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.2,
        "cacheWrite": 2.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-06-30"
    },
    {
      "id": "anthropic/claude-sonnet-5.5",
      "orId": "anthropic/claude-sonnet-5.5",
      "vendor": "anthropic",
      "model": "claude-sonnet-5.5",
      "name": "Anthropic: Claude Sonnet 5.5",
      "blurb": "Claude Sonnet 5.5 is Anthropic's Sonnet-class model for well-scoped everyday work, succeeding Claude Sonnet 5 as a direct upgrade. It is especially strong at building features, fixing bugs, and producing...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.2,
        "cacheWrite": 5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-28",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 166.56,
        "accessibility": "API access",
        "gpqa": 0.956,
        "gpqaStderr": 0.0137,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.465,
        "simpleQAStderr": 0.0158
      }
    },
    {
      "id": "anthropic/claude-sonnet-5.5:batch",
      "orId": "anthropic/claude-sonnet-5.5:batch",
      "vendor": "anthropic",
      "model": "claude-sonnet-5.5:batch",
      "name": "Anthropic: Claude Sonnet 5.5 (batch)",
      "blurb": "Claude Sonnet 5.5 is Anthropic's Sonnet-class model for well-scoped everyday work, succeeding Claude Sonnet 5 as a direct upgrade. It is especially strong at building features, fixing bugs, and producing...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.1,
        "cacheWrite": 2.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-28"
    },
    {
      "id": "baidu/ernie-5.5-flash-preview:free",
      "orId": "baidu/ernie-5.5-flash-preview:free",
      "vendor": "baidu",
      "model": "ernie-5.5-flash-preview:free",
      "name": "Baidu: ERNIE-5.5-Flash-Preview (free)",
      "blurb": "ERNIE 5.5 Flash Preview is the latest foundation model in the ERNIE family, with comprehensive improvements in coding, agentic capabilities, knowledge work, and deep research. It has 280 billion total...",
      "contextWindow": 1024000,
      "maxOutput": 921600,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-11-10",
      "openWeights": {
        "huggingFace": "baidu/ERNIE-5.5-Flash-Preview"
      }
    },
    {
      "id": "baidu/ernie-light-dt-1.0-intl",
      "orId": "baidu/ernie-light-dt-1.0-intl",
      "vendor": "baidu",
      "model": "ernie-light-dt-1.0-intl",
      "name": "Baidu: ERNIE-Light-DT-1.0",
      "blurb": "ERNIE-Light-DT-1.0 is based on open-source GLM models and includes research findings from Baidu Search's Lightning large model. It features Baidu Sparse Attention (BSA), a unique sparse attention mechanism developed by...",
      "contextWindow": 1024000,
      "maxOutput": 131072,
      "cost": {
        "input": 2.26,
        "output": 7.88,
        "cacheRead": 0.226
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-11-10"
    },
    {
      "id": "bytedance-seed/seed-2-1-turbo",
      "orId": "bytedance-seed/seed-2-1-turbo",
      "vendor": "bytedance-seed",
      "model": "seed-2-1-turbo",
      "name": "ByteDance Seed: Seed 2.1 Turbo",
      "blurb": "Seed 2.1 Turbo is a multimodal model from ByteDance Seed for coding and long-horizon agent workflows. It is suited for end-to-end software delivery, multi-step task execution, and understanding visual and...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 1,
        "output": 5
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-08-12"
    },
    {
      "id": "bytedance-seed/seed-2.0-lite",
      "orId": "bytedance-seed/seed-2.0-lite",
      "vendor": "bytedance-seed",
      "model": "seed-2.0-lite",
      "name": "ByteDance Seed: Seed-2.0-Lite",
      "blurb": "Seed-2.0-Lite is a versatile, cost‑efficient enterprise workhorse that delivers strong multimodal and agent capabilities while offering noticeably lower latency, making it a practical default choice for most production workloads across...",
      "contextWindow": 262144,
      "maxOutput": 131072,
      "cost": {
        "input": 0.5,
        "output": 4,
        "tiers": [
          {
            "minPromptTokens": 128000,
            "input": 1,
            "output": 8
          }
        ]
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-03-10"
    },
    {
      "id": "bytedance-seed/seed-2.0-mini",
      "orId": "bytedance-seed/seed-2.0-mini",
      "vendor": "bytedance-seed",
      "model": "seed-2.0-mini",
      "name": "ByteDance Seed: Seed-2.0-Mini",
      "blurb": "Seed-2.0-mini targets latency-sensitive, high-concurrency, and cost-sensitive scenarios, emphasizing fast response and flexible inference deployment. It delivers performance comparable to ByteDance-Seed-1.6, supports 256k context, four reasoning effort modes (minimal/low/medium/h",
      "contextWindow": 262144,
      "maxOutput": 131072,
      "cost": {
        "input": 0.2,
        "output": 0.8,
        "tiers": [
          {
            "minPromptTokens": 128000,
            "input": 0.4,
            "output": 1.6
          }
        ]
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-02-26"
    },
    {
      "id": "bytedance/ui-tars-1.5-7b",
      "orId": "bytedance/ui-tars-1.5-7b",
      "vendor": "bytedance",
      "model": "ui-tars-1.5-7b",
      "name": "ByteDance: UI-TARS 7B",
      "blurb": "UI-TARS-1.5 is a multimodal vision-language agent optimized for GUI-based environments, including desktop interfaces, web browsers, mobile systems, and games. Built by ByteDance, it builds upon the UI-TARS framework with reinforcement...",
      "contextWindow": 128000,
      "maxOutput": 2048,
      "cost": {
        "input": 0.2,
        "output": 0.4,
        "cacheRead": 0.2
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-07-22",
      "openWeights": {
        "huggingFace": "ByteDance-Seed/UI-TARS-1.5-7B"
      }
    },
    {
      "id": "cognitivecomputations/dolphin-mistral-24b-venice-edition",
      "orId": "cognitivecomputations/dolphin-mistral-24b-venice-edition",
      "vendor": "cognitivecomputations",
      "model": "dolphin-mistral-24b-venice-edition",
      "name": "Venice: Uncensored",
      "blurb": "Venice Uncensored Dolphin Mistral 24B Venice Edition is a fine-tuned variant of Mistral-Small-24B-Instruct-2501, developed by dphn.ai in collaboration with Venice.ai. This model is designed as an “uncensored” instruct-tuned LLM, preserving...",
      "contextWindow": 128000,
      "maxOutput": 8192,
      "cost": {
        "input": 0.4,
        "output": 1.8
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-04-30",
      "released": "2025-07-09",
      "openWeights": {
        "huggingFace": "cognitivecomputations/Dolphin-Mistral-24B-Venice-Edition"
      }
    },
    {
      "id": "cohere/command-a",
      "orId": "cohere/command-a",
      "vendor": "cohere",
      "model": "command-a",
      "name": "Cohere: Command A",
      "blurb": "Command A is an open-weights 111B parameter model with a 256k context window focused on delivering great performance across agentic, multilingual, and coding use cases. Compared to other leading proprietary...",
      "contextWindow": 256000,
      "maxOutput": 8192,
      "cost": {
        "input": 5,
        "output": 20
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-08-31",
      "released": "2025-03-13",
      "openWeights": {
        "huggingFace": "CohereForAI/c4ai-command-a-03-2025"
      }
    },
    {
      "id": "cohere/command-a-plus",
      "orId": "cohere/command-a-plus",
      "vendor": "cohere",
      "model": "command-a-plus",
      "name": "Cohere: Command A+",
      "blurb": "Command A+ is Cohere's flagship model for enterprise agentic workflows. It accepts text and image inputs with a 192K context window, supports native tool calling with strict tool schemas, structured...",
      "contextWindow": 192000,
      "maxOutput": 64000,
      "cost": {
        "input": 0.6,
        "output": 3,
        "cacheRead": 0.3
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-09-22",
      "openWeights": {
        "huggingFace": "CohereLabs/command-a-plus-05-2026-bf16"
      }
    },
    {
      "id": "cohere/command-r-08-2024",
      "orId": "cohere/command-r-08-2024",
      "vendor": "cohere",
      "model": "command-r-08-2024",
      "name": "Cohere: Command R (08-2024)",
      "blurb": "command-r-08-2024 is an update of the [Command R](/models/cohere/command-r) with improved performance for multilingual retrieval-augmented generation (RAG) and tool use. More broadly, it is better at math, code and reasoning and...",
      "contextWindow": 128000,
      "maxOutput": 4000,
      "cost": {
        "input": 0.3,
        "output": 1.2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-03-31",
      "released": "2024-08-30",
      "openWeights": {
        "huggingFace": "CohereLabs/c4ai-command-r-08-2024"
      }
    },
    {
      "id": "cohere/command-r-plus-08-2024",
      "orId": "cohere/command-r-plus-08-2024",
      "vendor": "cohere",
      "model": "command-r-plus-08-2024",
      "name": "Cohere: Command R+ (08-2024)",
      "blurb": "command-r-plus-08-2024 is an update of the [Command R+](/models/cohere/command-r-plus) with roughly 50% higher throughput and 25% lower latencies as compared to the previous Command R+ version, while keeping the hardware footprint...",
      "contextWindow": 128000,
      "maxOutput": 4000,
      "cost": {
        "input": 5,
        "output": 20
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-03-31",
      "released": "2024-08-30",
      "openWeights": {
        "huggingFace": "CohereLabs/c4ai-command-r-plus-08-2024"
      }
    },
    {
      "id": "cohere/command-r7b-12-2024",
      "orId": "cohere/command-r7b-12-2024",
      "vendor": "cohere",
      "model": "command-r7b-12-2024",
      "name": "Cohere: Command R7B (12-2024)",
      "blurb": "Command R7B (12-2024) is a small, fast update of the Command R+ model, delivered in December 2024. It excels at RAG, tool use, agents, and similar tasks requiring complex reasoning...",
      "contextWindow": 128000,
      "maxOutput": 4000,
      "cost": {
        "input": 0.075,
        "output": 0.3
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-08-31",
      "released": "2024-12-14",
      "openWeights": {
        "huggingFace": "CohereLabs/c4ai-command-r7b-12-2024"
      }
    },
    {
      "id": "cohere/north-mini-code:free",
      "orId": "cohere/north-mini-code:free",
      "vendor": "cohere",
      "model": "north-mini-code:free",
      "name": "Cohere: North Mini Code (free)",
      "blurb": "North Mini Code is Cohere's first agentic coding model and the debut of its North family. A sparse mixture-of-experts model with 30B total parameters and 3B active, it is optimized...",
      "contextWindow": 256000,
      "maxOutput": 64000,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-06-17",
      "openWeights": {
        "huggingFace": "CohereLabs/North-Mini-Code-1.0"
      }
    },
    {
      "id": "deepseek/deepseek-chat",
      "orId": "deepseek/deepseek-chat",
      "vendor": "deepseek",
      "model": "deepseek-chat",
      "name": "DeepSeek: DeepSeek V3",
      "blurb": "DeepSeek-V3 is the latest model from the DeepSeek team, building upon the instruction following and coding abilities of the previous versions. Pre-trained on nearly 15 trillion tokens, the reported evaluations...",
      "contextWindow": 163840,
      "maxOutput": 16000,
      "cost": {
        "input": 0.5148,
        "output": 2.0574
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-07-31",
      "released": "2024-12-26",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V3"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 131.7,
        "accessibility": "Open weights (restricted use)",
        "gpqa": 0.712,
        "gpqaStderr": 0.0323,
        "aime": 0.489,
        "aimeStderr": 0.0754
      }
    },
    {
      "id": "deepseek/deepseek-chat-v3-0324",
      "orId": "deepseek/deepseek-chat-v3-0324",
      "vendor": "deepseek",
      "model": "deepseek-chat-v3-0324",
      "name": "DeepSeek: DeepSeek V3 0324",
      "blurb": "DeepSeek V3, a 685B-parameter, mixture-of-experts model, is the latest iteration of the flagship chat model family from the DeepSeek team. It succeeds the [DeepSeek V3](/deepseek/deepseek-chat-v3) model and performs really well...",
      "contextWindow": 163840,
      "maxOutput": 115200,
      "cost": {
        "input": 0.58,
        "output": 2.28,
        "cacheRead": 0.22
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-07-31",
      "released": "2025-03-24",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V3-0324"
      },
      "bench": {
        "source": "epoch",
        "gpqa": 0.676,
        "gpqaStderr": 0.0268,
        "mathLevel5": 0.755,
        "mathLevel5Stderr": 0.0086,
        "aime": 0.378,
        "aimeStderr": 0.0624
      }
    },
    {
      "id": "deepseek/deepseek-chat-v3.1",
      "orId": "deepseek/deepseek-chat-v3.1",
      "vendor": "deepseek",
      "model": "deepseek-chat-v3.1",
      "name": "DeepSeek: DeepSeek V3.1",
      "blurb": "DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) that supports both thinking and non-thinking modes via prompt templates. It extends the DeepSeek-V3 base with a two-phase long-context...",
      "contextWindow": 163840,
      "maxOutput": 147456,
      "cost": {
        "input": 0.5,
        "output": 1.9,
        "cacheRead": 0.26
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-08-21",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V3.1"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 140.85,
        "accessibility": "Open weights (unrestricted)"
      }
    },
    {
      "id": "deepseek/deepseek-r1-0528",
      "orId": "deepseek/deepseek-r1-0528",
      "vendor": "deepseek",
      "model": "deepseek-r1-0528",
      "name": "DeepSeek: R1 0528",
      "blurb": "May 28th update to the [original DeepSeek R1](/deepseek/deepseek-r1) Performance on par with [OpenAI o1](/openai/o1), but open-sourced and with fully open reasoning tokens. It's 671B parameters in size, with 37B active...",
      "contextWindow": 163840,
      "maxOutput": 32768,
      "cost": {
        "input": 1,
        "output": 4.3,
        "cacheRead": 0.7
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-05-28",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-R1-0528"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "gpqa": 0.763,
        "gpqaStderr": 0.024,
        "mathLevel5": 0.966,
        "mathLevel5Stderr": 0.0042,
        "aime": 0.664,
        "aimeStderr": 0.0491
      }
    },
    {
      "id": "deepseek/deepseek-v3.1-terminus",
      "orId": "deepseek/deepseek-v3.1-terminus",
      "vendor": "deepseek",
      "model": "deepseek-v3.1-terminus",
      "name": "DeepSeek: DeepSeek V3.1 Terminus",
      "blurb": "DeepSeek-V3.1 Terminus is an update to [DeepSeek V3.1](/deepseek/deepseek-chat-v3.1) that maintains the model's original capabilities while addressing issues reported by users, including language consistency and agent capabilities, further optimizing the model's...",
      "contextWindow": 163840,
      "maxOutput": 147456,
      "cost": {
        "input": 0.54,
        "output": 2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-09-22",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V3.1-Terminus"
      }
    },
    {
      "id": "deepseek/deepseek-v3.2",
      "orId": "deepseek/deepseek-v3.2",
      "vendor": "deepseek",
      "model": "deepseek-v3.2",
      "name": "DeepSeek: DeepSeek V3.2",
      "blurb": "DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...",
      "contextWindow": 163840,
      "maxOutput": 147456,
      "cost": {
        "input": 0.52,
        "output": 0.76,
        "cacheRead": 0.26
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2025-12-01",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V3.2"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 146.25,
        "accessibility": "Open weights (unrestricted)"
      }
    },
    {
      "id": "deepseek/deepseek-v3.2-exp",
      "orId": "deepseek/deepseek-v3.2-exp",
      "vendor": "deepseek",
      "model": "deepseek-v3.2-exp",
      "name": "DeepSeek: DeepSeek V3.2 Exp",
      "blurb": "DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...",
      "contextWindow": 163840,
      "maxOutput": 147456,
      "cost": {
        "input": 0.54,
        "output": 0.82
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-07-31",
      "released": "2025-09-29",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V3.2-Exp"
      },
      "bench": {
        "source": "epoch",
        "eci": 144.38,
        "accessibility": "Open weights (unrestricted)"
      }
    },
    {
      "id": "deepseek/deepseek-v4-flash",
      "orId": "deepseek/deepseek-v4-flash",
      "vendor": "deepseek",
      "model": "deepseek-v4-flash",
      "name": "DeepSeek: DeepSeek V4 Flash 0423",
      "blurb": "DeepSeek V4 Flash is an efficiency-optimized Mixture-of-Experts model from DeepSeek with 284B total parameters and 13B activated parameters, supporting a 1M-token context window. It is designed for fast inference and...",
      "contextWindow": 1048576,
      "maxOutput": 393216,
      "cost": {
        "input": 0.18,
        "output": 0.36,
        "cacheRead": 0.036
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-04-24",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V4-Flash"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 146.85,
        "accessibility": "Open weights (unrestricted)"
      }
    },
    {
      "id": "deepseek/deepseek-v4-flash-0731",
      "orId": "deepseek/deepseek-v4-flash-0731",
      "vendor": "deepseek",
      "model": "deepseek-v4-flash-0731",
      "name": "DeepSeek: DeepSeek V4 Flash 0731",
      "blurb": "DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows....",
      "contextWindow": 1048576,
      "maxOutput": 384000,
      "cost": {
        "input": 0.12,
        "output": 0.36,
        "cacheRead": 0.03
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-07-31",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V4-Flash-0731"
      },
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 154.41,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.91,
        "gpqaStderr": 0.0177,
        "aime": 0.944,
        "aimeStderr": 0.0209,
        "simpleQA": 0.336,
        "simpleQAStderr": 0.015
      }
    },
    {
      "id": "deepseek/deepseek-v4-flash-0731:batch",
      "orId": "deepseek/deepseek-v4-flash-0731:batch",
      "vendor": "deepseek",
      "model": "deepseek-v4-flash-0731:batch",
      "name": "DeepSeek: DeepSeek V4 Flash 0731 (batch)",
      "blurb": "DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows....",
      "contextWindow": 1048576,
      "maxOutput": 384000,
      "cost": {
        "input": 0.096,
        "output": 0.288,
        "cacheRead": 0.024
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-07-31",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V4-Flash-0731"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "deepseek/deepseek-v4-flash-vision-exp",
      "orId": "deepseek/deepseek-v4-flash-vision-exp",
      "vendor": "deepseek",
      "model": "deepseek-v4-flash-vision-exp",
      "name": "DeepSeek: DeepSeek V4 Flash Vision Exp",
      "blurb": "DeepSeek V4 Flash Vision Exp is an experimental vision-enabled version of DeepSeek V4 Flash 0731 from DeepSeek, adding image understanding while matching the base model on text capabilities including agents,...",
      "contextWindow": 1048576,
      "maxOutput": 943717,
      "cost": {
        "input": 0.88,
        "output": 2.64,
        "cacheRead": 0.028
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-08-21",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V4-Flash-Vision-Exp"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "deepseek/deepseek-v4-pro",
      "orId": "deepseek/deepseek-v4-pro",
      "vendor": "deepseek",
      "model": "deepseek-v4-pro",
      "name": "DeepSeek: DeepSeek V4 Pro 0423",
      "blurb": "DeepSeek V4 Pro is a large-scale Mixture-of-Experts model from DeepSeek with 1.6T total parameters and 49B activated parameters, supporting a 1M-token context window. It is designed for advanced reasoning, coding,...",
      "contextWindow": 1048576,
      "maxOutput": 393216,
      "cost": {
        "input": 1.8,
        "output": 3,
        "cacheRead": 0.5
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-04-24",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V4-Pro"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 150.17,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.909,
        "gpqaStderr": 0.0205,
        "sweBenchVerified": 0.776,
        "sweBenchVerifiedStderr": 0.019,
        "aime": 0.967,
        "aimeStderr": 0.02,
        "simpleQA": 0.47,
        "simpleQAStderr": 0.0158
      }
    },
    {
      "id": "deepseek/deepseek-v4-pro-0813",
      "orId": "deepseek/deepseek-v4-pro-0813",
      "vendor": "deepseek",
      "model": "deepseek-v4-pro-0813",
      "name": "DeepSeek: DeepSeek V4 Pro 0813",
      "blurb": "DeepSeek V4 Pro 0813 is a large-scale mixture-of-experts model from DeepSeek. This is the GA release of DeepSeek V4 Pro.",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.222,
        "output": 8.88,
        "cacheRead": 0.194
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-08-12",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V4-Pro-0813"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 155.94,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.917,
        "gpqaStderr": 0.0152,
        "aime": 0.986,
        "aimeStderr": 0.0099,
        "simpleQA": 0.529,
        "simpleQAStderr": 0.0158
      }
    },
    {
      "id": "deepseek/deepseek-v4.1-flash",
      "orId": "deepseek/deepseek-v4.1-flash",
      "vendor": "deepseek",
      "model": "deepseek-v4.1-flash",
      "name": "DeepSeek: DeepSeek V4.1 Flash",
      "blurb": "DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...",
      "contextWindow": 1048576,
      "maxOutput": 393216,
      "cost": {
        "input": 0.28,
        "output": 0.84,
        "cacheRead": 0.0084
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-10",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V4.1-Flash"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 156.7,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.898,
        "gpqaStderr": 0.0173,
        "aime": 0.983,
        "aimeStderr": 0.0064,
        "simpleQA": 0.429,
        "simpleQAStderr": 0.0157
      }
    },
    {
      "id": "deepseek/deepseek-v4.1-flash-1017",
      "orId": "deepseek/deepseek-v4.1-flash-1017",
      "vendor": "deepseek",
      "model": "deepseek-v4.1-flash-1017",
      "name": "DeepSeek: DeepSeek V4.1 Flash 1017",
      "blurb": "DeepSeek V4.1 Flash 1017 is the October 17, 2026 snapshot of DeepSeek V4.1 Flash, served by DeepSeek under the `deepseek-flash` API model name. V4.1 Flash is a sparse mixture-of-experts model...",
      "contextWindow": 1048576,
      "maxOutput": 393216,
      "cost": {
        "input": 0.6,
        "output": 2.4,
        "cacheRead": 0.012,
        "tiers": [
          {
            "input": 0.3,
            "output": 1.2,
            "cacheRead": 0.006
          },
          {
            "input": 0.3,
            "output": 1.2,
            "cacheRead": 0.006
          },
          {
            "input": 0.6,
            "output": 2.4,
            "cacheRead": 0.012
          },
          {
            "input": 0.3,
            "output": 1.2,
            "cacheRead": 0.006
          },
          {
            "input": 0.6,
            "output": 2.4,
            "cacheRead": 0.012
          },
          {
            "input": 0.3,
            "output": 1.2,
            "cacheRead": 0.006
          }
        ]
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-10-17",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V4.1-Flash-1017"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 159.41,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.932,
        "gpqaStderr": 0.0157,
        "aime": 0.994,
        "aimeStderr": 0.0039,
        "simpleQA": 0.425,
        "simpleQAStderr": 0.0156
      }
    },
    {
      "id": "deepseek/deepseek-v4.1-flash-1017:batch",
      "orId": "deepseek/deepseek-v4.1-flash-1017:batch",
      "vendor": "deepseek",
      "model": "deepseek-v4.1-flash-1017:batch",
      "name": "DeepSeek: DeepSeek V4.1 Flash 1017 (batch)",
      "blurb": "DeepSeek V4.1 Flash 1017 is the October 17, 2026 snapshot of DeepSeek V4.1 Flash, served by DeepSeek under the `deepseek-flash` API model name. V4.1 Flash is a sparse mixture-of-experts model...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.074,
        "output": 0.65,
        "cacheRead": 0.006
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-10-17",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V4.1-Flash-1017"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "deepseek/deepseek-v4.1-flash:batch",
      "orId": "deepseek/deepseek-v4.1-flash:batch",
      "vendor": "deepseek",
      "model": "deepseek-v4.1-flash:batch",
      "name": "DeepSeek: DeepSeek V4.1 Flash (batch)",
      "blurb": "DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0.224,
        "output": 0.672,
        "cacheRead": 0.00672
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-10",
      "openWeights": {
        "huggingFace": "deepseek-ai/DeepSeek-V4.1-Flash"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "dots-studio/dots-3-note-preview:free",
      "orId": "dots-studio/dots-3-note-preview:free",
      "vendor": "dots-studio",
      "model": "dots-3-note-preview:free",
      "name": "Dots Studio: Dots3-Note Preview (free)",
      "blurb": "Dots3-Note Preview is an open-weight mixture-of-experts model from Dots Studio, with 16B active parameters out of 280B total. It is the lightest model in the Dots 3 family and is...",
      "contextWindow": 512000,
      "maxOutput": 460800,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-08-14",
      "retiresAt": "2026-12-31",
      "openWeights": {
        "huggingFace": "dots-studio/dots3-note-prev"
      }
    },
    {
      "id": "elorian/glimpse",
      "orId": "elorian/glimpse",
      "vendor": "elorian",
      "model": "glimpse",
      "name": "Elorian: Glimpse",
      "blurb": "Glimpse is a multimodal visual reasoning model from Elorian, built for visual understanding and complex visual reasoning in the physical world. It reasons directly over images and video rather than...",
      "contextWindow": 262144,
      "maxOutput": 256000,
      "cost": {
        "input": 1.2,
        "output": 6,
        "cacheRead": 0.12
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-10-17"
    },
    {
      "id": "fireworks/ember-1",
      "orId": "fireworks/ember-1",
      "vendor": "fireworks",
      "model": "ember-1",
      "name": "Fireworks: Ember-1",
      "blurb": "Ember-1 is a specialized reasoning model from Fireworks Research, built on Kimi K3. It is designed to make every token go further: it produces shorter reasoning traces, using roughly 40%...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 6,
        "output": 30,
        "cacheRead": 0.6
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-09-24"
    },
    {
      "id": "google/gemini-2.5-flash",
      "orId": "google/gemini-2.5-flash",
      "vendor": "google",
      "model": "gemini-2.5-flash",
      "name": "Google: Gemini 2.5 Flash",
      "blurb": "Gemini 2.5 Flash is Google's state-of-the-art workhorse model, specifically designed for advanced reasoning, coding, mathematics, and scientific tasks. It includes built-in \"thinking\" capabilities, enabling it to provide responses with greater...",
      "contextWindow": 1048576,
      "maxOutput": 65535,
      "cost": {
        "input": 0.6,
        "output": 5,
        "cacheRead": 0.06,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 2
      },
      "input": [
        "file",
        "image",
        "text",
        "audio",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-06-17",
      "bench": {
        "source": "epoch",
        "frontierMath": 0.048,
        "frontierMathStderr": 0.0127
      }
    },
    {
      "id": "google/gemini-2.5-flash-image",
      "orId": "google/gemini-2.5-flash-image",
      "vendor": "google",
      "model": "gemini-2.5-flash-image",
      "name": "Google: Nano Banana (Gemini 2.5 Flash Image)",
      "blurb": "Gemini 2.5 Flash Image, a.k.a. \"Nano Banana,\" is now generally available. It is a state of the art image generation model with contextual understanding. It is capable of image generation,...",
      "contextWindow": 32768,
      "maxOutput": 8192,
      "cost": {
        "input": 0.6,
        "output": 5,
        "cacheRead": 0.06,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 2
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "image",
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-10-07"
    },
    {
      "id": "google/gemini-2.5-flash-lite",
      "orId": "google/gemini-2.5-flash-lite",
      "vendor": "google",
      "model": "gemini-2.5-flash-lite",
      "name": "Google: Gemini 2.5 Flash Lite",
      "blurb": "Gemini 2.5 Flash-Lite is a lightweight reasoning model in the Gemini 2.5 family, optimized for ultra-low latency and cost efficiency. It offers improved throughput, faster token generation, and better performance...",
      "contextWindow": 1048576,
      "maxOutput": 65535,
      "cost": {
        "input": 0.2,
        "output": 0.8,
        "cacheRead": 0.02,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 0.6
      },
      "input": [
        "text",
        "image",
        "file",
        "audio",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-07-22"
    },
    {
      "id": "google/gemini-2.5-pro",
      "orId": "google/gemini-2.5-pro",
      "vendor": "google",
      "model": "gemini-2.5-pro",
      "name": "Google: Gemini 2.5 Pro",
      "blurb": "Gemini 2.5 Pro is Google’s state-of-the-art AI model designed for advanced reasoning, coding, mathematics, and scientific tasks. It employs “thinking” capabilities, enabling it to reason through responses with enhanced accuracy...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 2.5,
        "output": 20,
        "cacheRead": 0.25,
        "cacheWrite": 0.75,
        "webSearch": 0.028,
        "audioIn": 2.5,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 5,
            "output": 30,
            "cacheRead": 0.5
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file",
        "audio",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-06-17",
      "bench": {
        "source": "epoch",
        "gpqa": 0.853,
        "gpqaStderr": 0.0212,
        "sweBenchVerified": 0.576,
        "sweBenchVerifiedStderr": 0.0225,
        "frontierMath": 0.141,
        "frontierMathStderr": 0.0205,
        "aime": 0.842,
        "aimeStderr": 0.0481
      }
    },
    {
      "id": "google/gemini-2.5-pro-preview",
      "orId": "google/gemini-2.5-pro-preview",
      "vendor": "google",
      "model": "gemini-2.5-pro-preview",
      "name": "Google: Gemini 2.5 Pro Preview 06-05",
      "blurb": "Gemini 2.5 Pro is Google’s state-of-the-art AI model designed for advanced reasoning, coding, mathematics, and scientific tasks. It employs “thinking” capabilities, enabling it to reason through responses with enhanced accuracy...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 2.5,
        "output": 20,
        "cacheRead": 0.25,
        "cacheWrite": 0.75,
        "webSearch": 0.028,
        "audioIn": 2.5,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 5,
            "output": 30,
            "cacheRead": 0.5
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-06-05",
      "bench": {
        "source": "epoch",
        "gpqa": 0.848,
        "gpqaStderr": 0.0255,
        "frontierMath": 0.103,
        "frontierMathStderr": 0.0179
      }
    },
    {
      "id": "google/gemini-3-flash-preview",
      "orId": "google/gemini-3-flash-preview",
      "vendor": "google",
      "model": "gemini-3-flash-preview",
      "name": "Google: Gemini 3 Flash Preview",
      "blurb": "Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance. It delivers near Pro level reasoning and tool...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 1,
        "output": 6,
        "cacheRead": 0.1,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 2
      },
      "input": [
        "text",
        "image",
        "file",
        "audio",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-12-17",
      "bench": {
        "source": "epoch",
        "gpqa": 0.894,
        "gpqaStderr": 0.0219,
        "sweBenchVerified": 0.754,
        "sweBenchVerifiedStderr": 0.0196,
        "frontierMath": 0.356,
        "frontierMathStderr": 0.0282,
        "aime": 0.956,
        "aimeStderr": 0.0311,
        "simpleQA": 0.668,
        "simpleQAStderr": 0.0149
      }
    },
    {
      "id": "google/gemini-3-flash-preview:batch",
      "orId": "google/gemini-3-flash-preview:batch",
      "vendor": "google",
      "model": "gemini-3-flash-preview:batch",
      "name": "Google: Gemini 3 Flash Preview (batch)",
      "blurb": "Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance. It delivers near Pro level reasoning and tool...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0.5,
        "output": 3,
        "webSearch": 0.028,
        "audioIn": 1
      },
      "input": [
        "text",
        "image",
        "file",
        "audio",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-12-17"
    },
    {
      "id": "google/gemini-3-pro-image",
      "orId": "google/gemini-3-pro-image",
      "vendor": "google",
      "model": "gemini-3-pro-image",
      "name": "Google: Nano Banana Pro (Gemini 3 Pro Image)",
      "blurb": "Nano Banana Pro is Google’s most advanced image-generation and editing model, built on Gemini 3 Pro. It extends the original Nano Banana with significantly improved multimodal reasoning, real-world grounding, and...",
      "contextWindow": 131072,
      "maxOutput": 32768,
      "cost": {
        "input": 4,
        "output": 24,
        "cacheRead": 0.4,
        "cacheWrite": 0.75,
        "webSearch": 0.028,
        "audioIn": 4
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "image",
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-06-18"
    },
    {
      "id": "google/gemini-3-pro-image-preview",
      "orId": "google/gemini-3-pro-image-preview",
      "vendor": "google",
      "model": "gemini-3-pro-image-preview",
      "name": "Google: Nano Banana Pro (Gemini 3 Pro Image Preview)",
      "blurb": "Nano Banana Pro is Google’s most advanced image-generation and editing model, built on Gemini 3 Pro. It extends the original Nano Banana with significantly improved multimodal reasoning, real-world grounding, and...",
      "contextWindow": 65536,
      "maxOutput": 32768,
      "cost": {
        "input": 4,
        "output": 24,
        "cacheRead": 0.4,
        "cacheWrite": 0.75,
        "webSearch": 0.028,
        "audioIn": 4
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "image",
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": false,
      "released": "2025-11-20"
    },
    {
      "id": "google/gemini-3.1-flash-image",
      "orId": "google/gemini-3.1-flash-image",
      "vendor": "google",
      "model": "gemini-3.1-flash-image",
      "name": "Google: Nano Banana 2 (Gemini 3.1 Flash Image)",
      "blurb": "Gemini 3.1 Flash Image, a.k.a. \"Nano Banana 2,\" is Google’s latest state of the art image generation and editing model, delivering Pro-level visual quality at Flash speed. It combines advanced...",
      "contextWindow": 131072,
      "maxOutput": 32768,
      "cost": {
        "input": 1,
        "output": 6,
        "webSearch": 0.028
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "image",
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "minimal",
          "none"
        ],
        "defaultEffort": "minimal"
      },
      "tools": false,
      "released": "2026-06-18",
      "regions": [
        "us"
      ]
    },
    {
      "id": "google/gemini-3.1-flash-image-preview",
      "orId": "google/gemini-3.1-flash-image-preview",
      "vendor": "google",
      "model": "gemini-3.1-flash-image-preview",
      "name": "Google: Nano Banana 2 (Gemini 3.1 Flash Image Preview)",
      "blurb": "Gemini 3.1 Flash Image Preview, a.k.a. \"Nano Banana 2,\" is Google’s latest state of the art image generation and editing model, delivering Pro-level visual quality at Flash speed. It combines...",
      "contextWindow": 65536,
      "maxOutput": 58982,
      "cost": {
        "input": 1,
        "output": 6,
        "webSearch": 0.028
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "image",
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "minimal",
          "none"
        ],
        "defaultEffort": "minimal"
      },
      "tools": false,
      "released": "2026-02-26"
    },
    {
      "id": "google/gemini-3.1-flash-lite",
      "orId": "google/gemini-3.1-flash-lite",
      "vendor": "google",
      "model": "gemini-3.1-flash-lite",
      "name": "Google: Gemini 3.1 Flash Lite",
      "blurb": "Gemini 3.1 Flash Lite is Google’s GA high-efficiency multimodal model optimized for low-latency, high-volume workloads. It supports text, image, video, audio, and PDF inputs, and is designed for lightweight agentic...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0.5,
        "output": 3,
        "cacheRead": 0.05,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 1
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "minimal"
      },
      "tools": true,
      "released": "2026-05-07",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 145.01,
        "accessibility": "API access",
        "gpqa": 0.818,
        "gpqaStderr": 0.0275,
        "aime": 0.8,
        "aimeStderr": 0.0603
      }
    },
    {
      "id": "google/gemini-3.1-flash-lite-image",
      "orId": "google/gemini-3.1-flash-lite-image",
      "vendor": "google",
      "model": "gemini-3.1-flash-lite-image",
      "name": "Google: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)",
      "blurb": "Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) is Google's fastest, most cost-efficient Gemini image model, built for high-velocity developer pipelines and rapid-fire visual exploration. It delivers text-to-image generation...",
      "contextWindow": 65536,
      "maxOutput": 58982,
      "cost": {
        "input": 0.5,
        "output": 3,
        "webSearch": 0.028
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "image",
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "minimal",
          "none"
        ],
        "defaultEffort": "minimal"
      },
      "tools": false,
      "knowledgeCutoff": "2025-01-01",
      "released": "2026-06-30"
    },
    {
      "id": "google/gemini-3.1-flash-lite-preview",
      "orId": "google/gemini-3.1-flash-lite-preview",
      "vendor": "google",
      "model": "gemini-3.1-flash-lite-preview",
      "name": "Google: Gemini 3.1 Flash Lite Preview",
      "blurb": "Gemini 3.1 Flash Lite Preview is Google's high-efficiency model optimized for high-volume use cases. It outperforms Gemini 2.5 Flash Lite on overall quality and approaches Gemini 2.5 Flash performance across...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0.5,
        "output": 3,
        "cacheRead": 0.05,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 1
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "minimal"
      },
      "tools": true,
      "released": "2026-03-03"
    },
    {
      "id": "google/gemini-3.1-flash-lite:batch",
      "orId": "google/gemini-3.1-flash-lite:batch",
      "vendor": "google",
      "model": "gemini-3.1-flash-lite:batch",
      "name": "Google: Gemini 3.1 Flash Lite (batch)",
      "blurb": "Gemini 3.1 Flash Lite is Google’s GA high-efficiency multimodal model optimized for low-latency, high-volume workloads. It supports text, image, video, audio, and PDF inputs, and is designed for lightweight agentic...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0.25,
        "output": 1.5,
        "cacheRead": 0.025,
        "webSearch": 0.028,
        "audioIn": 0.5
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "minimal"
      },
      "tools": true,
      "released": "2026-05-07"
    },
    {
      "id": "google/gemini-3.1-pro-preview",
      "orId": "google/gemini-3.1-pro-preview",
      "vendor": "google",
      "model": "gemini-3.1-pro-preview",
      "name": "Google: Gemini 3.1 Pro Preview",
      "blurb": "Gemini 3.1 Pro Preview is Google’s frontier reasoning model, delivering enhanced software engineering performance, improved agentic reliability, and more efficient token usage across complex workflows. Building on the multimodal foundation...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 4,
        "output": 24,
        "cacheRead": 0.4,
        "cacheWrite": 0.75,
        "webSearch": 0.028,
        "audioIn": 4,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 8,
            "output": 36,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-02-19",
      "bench": {
        "source": "epoch",
        "gpqa": 0.944,
        "gpqaStderr": 0.0163,
        "frontierMath": 0.369,
        "frontierMathStderr": 0.028,
        "aime": 0.956,
        "aimeStderr": 0.0311,
        "simpleQA": 0.735,
        "simpleQAStderr": 0.014
      }
    },
    {
      "id": "google/gemini-3.1-pro-preview-customtools",
      "orId": "google/gemini-3.1-pro-preview-customtools",
      "vendor": "google",
      "model": "gemini-3.1-pro-preview-customtools",
      "name": "Google: Gemini 3.1 Pro Preview Custom Tools",
      "blurb": "Gemini 3.1 Pro Preview Custom Tools is a variant of Gemini 3.1 Pro that improves tool selection behavior by preventing overuse of a general bash tool when more efficient third-party...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 4,
        "output": 24,
        "cacheRead": 0.4,
        "cacheWrite": 0.75,
        "webSearch": 0.028,
        "audioIn": 4,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 8,
            "output": 36,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "text",
        "audio",
        "image",
        "video",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-02-25",
      "bench": {
        "source": "epoch",
        "sweBenchVerified": 0.756,
        "sweBenchVerifiedStderr": 0.0195
      }
    },
    {
      "id": "google/gemini-3.1-pro-preview:batch",
      "orId": "google/gemini-3.1-pro-preview:batch",
      "vendor": "google",
      "model": "gemini-3.1-pro-preview:batch",
      "name": "Google: Gemini 3.1 Pro Preview (batch)",
      "blurb": "Gemini 3.1 Pro Preview is Google’s frontier reasoning model, delivering enhanced software engineering performance, improved agentic reliability, and more efficient token usage across complex workflows. Building on the multimodal foundation...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 2,
        "output": 12,
        "webSearch": 0.028,
        "audioIn": 2,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 4,
            "output": 18
          }
        ]
      },
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-02-19"
    },
    {
      "id": "google/gemini-3.5-flash",
      "orId": "google/gemini-3.5-flash",
      "vendor": "google",
      "model": "gemini-3.5-flash",
      "name": "Google: Gemini 3.5 Flash",
      "blurb": "Gemini 3.5 Flash is Google's high-efficiency multimodal model, bringing near-Pro level coding and reasoning at Flash-tier cost and speed. It is highly optimized for coding proficiency and parallel agentic execution...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 3,
        "output": 18,
        "cacheRead": 0.3,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 6
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-01-01",
      "released": "2026-05-19",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 155.28,
        "accessibility": "API access",
        "gpqa": 0.928,
        "gpqaStderr": 0.0164,
        "sweBenchVerified": 0.793,
        "sweBenchVerifiedStderr": 0.0184,
        "frontierMath": 0.39,
        "frontierMathStderr": 0.0287,
        "aime": 0.956,
        "aimeStderr": 0.0267,
        "simpleQA": 0.662,
        "simpleQAStderr": 0.015
      }
    },
    {
      "id": "google/gemini-3.5-flash-lite",
      "orId": "google/gemini-3.5-flash-lite",
      "vendor": "google",
      "model": "gemini-3.5-flash-lite",
      "name": "Google: Gemini 3.5 Flash Lite",
      "blurb": "Gemini 3.5 Flash Lite is a high-efficiency model from Google with upgraded agentic capabilities. It is suited for subagents that execute focused tasks within complex, multi-agent workflows.",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0.6,
        "output": 5,
        "cacheRead": 0.06,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 0.6
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "minimal"
      },
      "tools": true,
      "released": "2026-07-21",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 145.63,
        "accessibility": "API access",
        "gpqa": 0.833,
        "gpqaStderr": 0.0266,
        "aime": 0.711,
        "aimeStderr": 0.0683
      }
    },
    {
      "id": "google/gemini-3.5-flash-lite:batch",
      "orId": "google/gemini-3.5-flash-lite:batch",
      "vendor": "google",
      "model": "gemini-3.5-flash-lite:batch",
      "name": "Google: Gemini 3.5 Flash Lite (batch)",
      "blurb": "Gemini 3.5 Flash Lite is a high-efficiency model from Google with upgraded agentic capabilities. It is suited for subagents that execute focused tasks within complex, multi-agent workflows.",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0.3,
        "output": 2.5,
        "cacheRead": 0.03,
        "webSearch": 0.028,
        "audioIn": 0.3
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "minimal"
      },
      "tools": true,
      "released": "2026-07-21"
    },
    {
      "id": "google/gemini-3.5-flash:batch",
      "orId": "google/gemini-3.5-flash:batch",
      "vendor": "google",
      "model": "gemini-3.5-flash:batch",
      "name": "Google: Gemini 3.5 Flash (batch)",
      "blurb": "Gemini 3.5 Flash is Google's high-efficiency multimodal model, bringing near-Pro level coding and reasoning at Flash-tier cost and speed. It is highly optimized for coding proficiency and parallel agentic execution...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 1.5,
        "output": 9,
        "cacheRead": 0.15,
        "webSearch": 0.028,
        "audioIn": 3
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-01-01",
      "released": "2026-05-19"
    },
    {
      "id": "google/gemini-3.6-flash",
      "orId": "google/gemini-3.6-flash",
      "vendor": "google",
      "model": "gemini-3.6-flash",
      "name": "Google: Gemini 3.6 Flash",
      "blurb": "Gemini 3.6 Flash is a high-efficiency model from Google for coding, agentic workflows, and web and app development. It is designed to produce polished outputs with fewer unnecessary edits and...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 1.5,
        "output": 7.5,
        "cacheRead": 0.15,
        "cacheWrite": 0.083334,
        "webSearch": 0.028,
        "audioIn": 1.5
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-07-21",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 155.21,
        "accessibility": "API access",
        "gpqa": 0.941,
        "gpqaStderr": 0.014,
        "aime": 0.942,
        "aimeStderr": 0.0309,
        "simpleQA": 0.662,
        "simpleQAStderr": 0.015
      }
    },
    {
      "id": "google/gemini-3.6-flash:batch",
      "orId": "google/gemini-3.6-flash:batch",
      "vendor": "google",
      "model": "gemini-3.6-flash:batch",
      "name": "Google: Gemini 3.6 Flash (batch)",
      "blurb": "Gemini 3.6 Flash is a high-efficiency model from Google for coding, agentic workflows, and web and app development. It is designed to produce polished outputs with fewer unnecessary edits and...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0.75,
        "output": 3.75,
        "cacheRead": 0.075,
        "cacheWrite": 0.083334,
        "webSearch": 0.028,
        "audioIn": 0.75
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-07-21"
    },
    {
      "id": "google/gemini-3.7-flash",
      "orId": "google/gemini-3.7-flash",
      "vendor": "google",
      "model": "gemini-3.7-flash",
      "name": "Google: Gemini 3.7 Flash",
      "blurb": "Gemini 3.7 Flash is a multimodal model from Google for fast agentic workflows, coding, and complex multi-step reasoning. It is designed for tasks that require responsive performance and reliable multi-step...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 1.5,
        "output": 7.5,
        "cacheRead": 0.15,
        "cacheWrite": 0.083334,
        "webSearch": 0.028,
        "audioIn": 1.5
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-08-13",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 158.36,
        "accessibility": "API access",
        "gpqa": 0.948,
        "gpqaStderr": 0.0135,
        "aime": 0.972,
        "aimeStderr": 0.0186,
        "simpleQA": 0.692,
        "simpleQAStderr": 0.0146
      }
    },
    {
      "id": "google/gemini-3.7-flash:batch",
      "orId": "google/gemini-3.7-flash:batch",
      "vendor": "google",
      "model": "gemini-3.7-flash:batch",
      "name": "Google: Gemini 3.7 Flash (batch)",
      "blurb": "Gemini 3.7 Flash is a multimodal model from Google for fast agentic workflows, coding, and complex multi-step reasoning. It is designed for tasks that require responsive performance and reliable multi-step...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0.75,
        "output": 3.75,
        "cacheRead": 0.075,
        "cacheWrite": 0.083334,
        "webSearch": 0.028,
        "audioIn": 0.75
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-08-13"
    },
    {
      "id": "google/gemini-3.8-flash",
      "orId": "google/gemini-3.8-flash",
      "vendor": "google",
      "model": "gemini-3.8-flash",
      "name": "Google: Gemini 3.8 Flash",
      "blurb": "Gemini 3.8 Flash is Google's most intelligent Flash model with significant gains from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning.",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 1.65,
        "output": 8.25,
        "cacheRead": 0.165,
        "cacheWrite": 0.083334,
        "webSearch": 0.028,
        "audioIn": 1.65
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-02",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 158.79,
        "accessibility": "API access",
        "gpqa": 0.954,
        "gpqaStderr": 0.014,
        "aime": 0.989,
        "aimeStderr": 0.0087,
        "simpleQA": 0.697,
        "simpleQAStderr": 0.0145
      }
    },
    {
      "id": "google/gemini-3.8-flash:batch",
      "orId": "google/gemini-3.8-flash:batch",
      "vendor": "google",
      "model": "gemini-3.8-flash:batch",
      "name": "Google: Gemini 3.8 Flash (batch)",
      "blurb": "Gemini 3.8 Flash is Google's most intelligent Flash model with significant gains from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning.",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0.75,
        "output": 3.75,
        "cacheRead": 0.075,
        "cacheWrite": 0.083334,
        "webSearch": 0.028,
        "audioIn": 0.75
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-02"
    },
    {
      "id": "google/gemini-4-argon",
      "orId": "google/gemini-4-argon",
      "vendor": "google",
      "model": "gemini-4-argon",
      "name": "Google: Gemini 4 Argon",
      "blurb": "Gemini 4 Argon is Google's flagship reasoning model, built for complex software engineering, deep mathematical and scientific problem solving, and autonomous multi-step agents. Google trained it to keep context across...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.2,
        "cacheWrite": 0.375,
        "webSearch": 0.028,
        "audioIn": 4
      },
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-13",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 166.48,
        "accessibility": "Limited access",
        "gpqa": 0.964,
        "gpqaStderr": 0.0119,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.76,
        "simpleQAStderr": 0.0135
      }
    },
    {
      "id": "google/gemini-4-argon:batch",
      "orId": "google/gemini-4-argon:batch",
      "vendor": "google",
      "model": "gemini-4-argon:batch",
      "name": "Google: Gemini 4 Argon (batch)",
      "blurb": "Gemini 4 Argon is Google's flagship reasoning model, built for complex software engineering, deep mathematical and scientific problem solving, and autonomous multi-step agents. Google trained it to keep context across...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.1,
        "cacheWrite": 0.375,
        "webSearch": 0.028,
        "audioIn": 2
      },
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-13"
    },
    {
      "id": "google/gemini-4-carbon",
      "orId": "google/gemini-4-carbon",
      "vendor": "google",
      "model": "gemini-4-carbon",
      "name": "Google: Gemini 4 Carbon",
      "blurb": "Gemini 4 Carbon is a frontier Gemini 4 model from Google for coding, agentic and complex reasoning work. It takes text, image, audio, video and file input, supports long decode...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.2,
        "cacheWrite": 0.375,
        "webSearch": 0.028,
        "audioIn": 4
      },
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-11-10",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 169.76,
        "accessibility": "Unreleased",
        "gpqa": 0.965,
        "gpqaStderr": 0.0121,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.777,
        "simpleQAStderr": 0.0132
      }
    },
    {
      "id": "google/gemini-4-carbon:batch",
      "orId": "google/gemini-4-carbon:batch",
      "vendor": "google",
      "model": "gemini-4-carbon:batch",
      "name": "Google: Gemini 4 Carbon (batch)",
      "blurb": "Gemini 4 Carbon is a frontier Gemini 4 model from Google for coding, agentic and complex reasoning work. It takes text, image, audio, video and file input, supports long decode...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.1,
        "cacheWrite": 0.375,
        "webSearch": 0.028,
        "audioIn": 2
      },
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-11-10"
    },
    {
      "id": "google/gemini-4-hydrogen",
      "orId": "google/gemini-4-hydrogen",
      "vendor": "google",
      "model": "gemini-4-hydrogen",
      "name": "Google: Gemini 4 Hydrogen",
      "blurb": "Gemini 4 Hydrogen is the high-efficiency workhorse model of the Gemini 4 family from Google, built for high-throughput agentic workflows, coding and multi-model routing at lower cost and latency than...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 3,
        "output": 15,
        "cacheRead": 0.15,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 3
      },
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-11-12",
      "regions": [
        "eu",
        "us"
      ]
    },
    {
      "id": "google/gemini-4-hydrogen:batch",
      "orId": "google/gemini-4-hydrogen:batch",
      "vendor": "google",
      "model": "gemini-4-hydrogen:batch",
      "name": "Google: Gemini 4 Hydrogen (batch)",
      "blurb": "Gemini 4 Hydrogen is the high-efficiency workhorse model of the Gemini 4 family from Google, built for high-throughput agentic workflows, coding and multi-model routing at lower cost and latency than...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 1.5,
        "output": 7.5,
        "cacheRead": 0.075,
        "cacheWrite": 0.166666,
        "webSearch": 0.028,
        "audioIn": 1.5
      },
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-11-12"
    },
    {
      "id": "google/gemini-nano-banana-2.1",
      "orId": "google/gemini-nano-banana-2.1",
      "vendor": "google",
      "model": "gemini-nano-banana-2.1",
      "name": "Google: Nano Banana 2.1",
      "blurb": "Nano Banana 2.1 (Gemini Nano Banana 2.1) is Google's image generation and editing model on the Flash tier, succeeding Nano Banana 2 and Nano Banana Pro. It improves product recontextualization,...",
      "contextWindow": 65536,
      "maxOutput": 58982,
      "cost": {
        "input": 3,
        "output": 15,
        "webSearch": 0.028
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "image",
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "minimal",
          "none"
        ],
        "defaultEffort": "minimal"
      },
      "tools": true,
      "released": "2026-10-06"
    },
    {
      "id": "google/gemini-omni-1.1-flash",
      "orId": "google/gemini-omni-1.1-flash",
      "vendor": "google",
      "model": "gemini-omni-1.1-flash",
      "name": "Google: Gemini Omni 1.1 Flash",
      "blurb": "Gemini Omni 1.1 Flash is Google's native multimodal video generation and editing model, the generally available successor to Gemini Omni Flash Preview. It is suited for text-to-video, image-to-video, first and...",
      "contextWindow": 131072,
      "maxOutput": 65536,
      "cost": {
        "input": 3,
        "output": 18,
        "webSearch": 0.028
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text",
        "video"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "low"
        ],
        "defaultEffort": "low"
      },
      "tools": false,
      "released": "2026-11-04"
    },
    {
      "id": "google/gemini-omni-1.2-flash",
      "orId": "google/gemini-omni-1.2-flash",
      "vendor": "google",
      "model": "gemini-omni-1.2-flash",
      "name": "Google: Gemini Omni 1.2 Flash",
      "blurb": "Gemini Omni 1.2 Flash is Google's native multimodal video generation and editing model, the successor to Gemini Omni 1.1 Flash. It generates clips up to 15 seconds long with native...",
      "contextWindow": 131072,
      "maxOutput": 65536,
      "cost": {
        "input": 3,
        "output": 18,
        "webSearch": 0.028
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text",
        "video"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "low"
        ],
        "defaultEffort": "low"
      },
      "tools": false,
      "released": "2026-11-05"
    },
    {
      "id": "google/gemma-2-27b-it",
      "orId": "google/gemma-2-27b-it",
      "vendor": "google",
      "model": "gemma-2-27b-it",
      "name": "Google: Gemma 2 27B",
      "blurb": "Gemma 2 27B by Google is an open model built from the same research and technology used to create the [Gemini models](/models?q=gemini). Gemma models are well-suited for a variety of...",
      "contextWindow": 8192,
      "maxOutput": 2048,
      "cost": {
        "input": 1.3,
        "output": 1.3
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-06-30",
      "released": "2024-07-13",
      "openWeights": {
        "huggingFace": "google/gemma-2-27b-it"
      },
      "bench": {
        "source": "epoch",
        "eci": 118.59,
        "accessibility": "Open weights (restricted use)",
        "gpqa": 0.365,
        "gpqaStderr": 0.0239,
        "mathLevel5": 0.279,
        "mathLevel5Stderr": 0.0099,
        "aime": 0.014,
        "aimeStderr": 0.0091
      }
    },
    {
      "id": "google/gemma-3-12b-it",
      "orId": "google/gemma-3-12b-it",
      "vendor": "google",
      "model": "gemma-3-12b-it",
      "name": "Google: Gemma 3 12B",
      "blurb": "Gemma 3 introduces multimodality, supporting vision-language input and text outputs. It handles context windows up to 128k tokens, understands over 140 languages, and offers improved math, reasoning, and chat capabilities,...",
      "contextWindow": 131072,
      "maxOutput": 16384,
      "cost": {
        "input": 0.1,
        "output": 0.3
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-08-31",
      "released": "2025-03-13",
      "openWeights": {
        "huggingFace": "google/gemma-3-12b-it"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "gpqa": 0.395,
        "gpqaStderr": 0.0253,
        "aime": 0.167,
        "aimeStderr": 0.0424
      }
    },
    {
      "id": "google/gemma-3-27b-it",
      "orId": "google/gemma-3-27b-it",
      "vendor": "google",
      "model": "gemma-3-27b-it",
      "name": "Google: Gemma 3 27B",
      "blurb": "Gemma 3 introduces multimodality, supporting vision-language input and text outputs. It handles context windows up to 128k tokens, understands over 140 languages, and offers improved math, reasoning, and chat capabilities,...",
      "contextWindow": 131072,
      "maxOutput": 117964,
      "cost": {
        "input": 0.16,
        "output": 0.7,
        "cacheRead": 0.08
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-08-31",
      "released": "2025-03-12",
      "openWeights": {
        "huggingFace": "google/gemma-3-27b-it"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 127.51,
        "accessibility": "Open weights (restricted use)",
        "gpqa": 0.477,
        "gpqaStderr": 0.027,
        "mathLevel5": 0.74,
        "mathLevel5Stderr": 0.0112,
        "aime": 0.225,
        "aimeStderr": 0.0528
      }
    },
    {
      "id": "google/gemma-3-4b-it",
      "orId": "google/gemma-3-4b-it",
      "vendor": "google",
      "model": "gemma-3-4b-it",
      "name": "Google: Gemma 3 4B",
      "blurb": "Gemma 3 introduces multimodality, supporting vision-language input and text outputs. It handles context windows up to 128k tokens, understands over 140 languages, and offers improved math, reasoning, and chat capabilities,...",
      "contextWindow": 131072,
      "maxOutput": 16384,
      "cost": {
        "input": 0.1,
        "output": 0.2
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-08-31",
      "released": "2025-03-13",
      "openWeights": {
        "huggingFace": "google/gemma-3-4b-it"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "gpqa": 0.232,
        "gpqaStderr": 0.0187,
        "aime": 0.075,
        "aimeStderr": 0.0276
      }
    },
    {
      "id": "google/gemma-4-26b-a4b-it",
      "orId": "google/gemma-4-26b-a4b-it",
      "vendor": "google",
      "model": "gemma-4-26b-a4b-it",
      "name": "Google: Gemma 4 26B A4B",
      "blurb": "Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.18,
        "output": 0.6,
        "cacheRead": 0.1
      },
      "input": [
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-03",
      "openWeights": {
        "huggingFace": "google/gemma-4-26B-A4B-it"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 141.64,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.732,
        "gpqaStderr": 0.0315,
        "aime": 0.822,
        "aimeStderr": 0.0576
      }
    },
    {
      "id": "google/gemma-4-26b-a4b-it:free",
      "orId": "google/gemma-4-26b-a4b-it:free",
      "vendor": "google",
      "model": "gemma-4-26b-a4b-it:free",
      "name": "Google: Gemma 4 26B A4B  (free)",
      "blurb": "Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at...",
      "contextWindow": 262144,
      "maxOutput": 32768,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-03",
      "openWeights": {
        "huggingFace": "google/gemma-4-26B-A4B-it"
      }
    },
    {
      "id": "google/gemma-4-31b-it",
      "orId": "google/gemma-4-31b-it",
      "vendor": "google",
      "model": "gemma-4-31b-it",
      "name": "Google: Gemma 4 31B",
      "blurb": "Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function...",
      "contextWindow": 262144,
      "maxOutput": 16384,
      "cost": {
        "input": 0.4,
        "output": 0.8
      },
      "input": [
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-02",
      "openWeights": {
        "huggingFace": "google/gemma-4-31B-it"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 144.27,
        "accessibility": "Open weights (restricted use)",
        "gpqa": 0.758,
        "gpqaStderr": 0.0305,
        "aime": 0.733,
        "aimeStderr": 0.0667,
        "simpleQA": 0.104,
        "simpleQAStderr": 0.0097
      }
    },
    {
      "id": "google/gemma-4-31b-it:free",
      "orId": "google/gemma-4-31b-it:free",
      "vendor": "google",
      "model": "gemma-4-31b-it:free",
      "name": "Google: Gemma 4 31B (free)",
      "blurb": "Gemma 4 31B Instruct is Google DeepMind's 30.7B dense multimodal model supporting text and image input with text output. Features a 256K token context window, configurable thinking/reasoning mode, native function...",
      "contextWindow": 262144,
      "maxOutput": 32768,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-02",
      "openWeights": {
        "huggingFace": "google/gemma-4-31B-it"
      }
    },
    {
      "id": "google/lyria-3-clip-preview",
      "orId": "google/lyria-3-clip-preview",
      "vendor": "google",
      "model": "lyria-3-clip-preview",
      "name": "Google: Lyria 3 Clip Preview",
      "blurb": "30 second duration clips are priced at $0.04 per clip. Lyria 3 is Google's family of music generation models, available through the Gemini API. With Lyria 3, you can generate...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text",
        "audio"
      ],
      "tools": false,
      "released": "2026-03-30"
    },
    {
      "id": "google/lyria-3-pro-preview",
      "orId": "google/lyria-3-pro-preview",
      "vendor": "google",
      "model": "lyria-3-pro-preview",
      "name": "Google: Lyria 3 Pro Preview",
      "blurb": "Full-length songs are priced at $0.08 per song. Lyria 3 is Google's family of music generation models, available through the Gemini API. With Lyria 3, you can generate high-quality, 48kHz...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text",
        "audio"
      ],
      "tools": false,
      "released": "2026-03-30"
    },
    {
      "id": "gryphe/mythomax-l2-13b",
      "orId": "gryphe/mythomax-l2-13b",
      "vendor": "gryphe",
      "model": "mythomax-l2-13b",
      "name": "MythoMax 13B",
      "blurb": "One of the highest performing and most popular fine-tunes of Llama 2 13B, with rich descriptions and roleplay. #merge",
      "contextWindow": 8192,
      "maxOutput": 3686,
      "cost": {
        "input": 0.16,
        "output": 0.22
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2023-06-30",
      "released": "2023-07-02",
      "openWeights": {
        "huggingFace": "Gryphe/MythoMax-L2-13b"
      }
    },
    {
      "id": "ibm-granite/granite-4.2-8b",
      "orId": "ibm-granite/granite-4.2-8b",
      "vendor": "ibm-granite",
      "model": "granite-4.2-8b",
      "name": "IBM: Granite 4.2 8B",
      "blurb": "Granite 4.2 8B is a dense reasoning model from IBM. It is suited for mathematics, code generation, multilingual dialogue, and agentic workflows that need multi-step reasoning. It supports full, low-effort,...",
      "contextWindow": 131072,
      "maxOutput": 117964,
      "cost": {
        "input": 0.12,
        "output": 0.3,
        "cacheRead": 0.03
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-08-31",
      "openWeights": {
        "huggingFace": "ibm-granite/granite-4.2-8b"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "inclusionai/ling-3.0-flash",
      "orId": "inclusionai/ling-3.0-flash",
      "vendor": "inclusionai",
      "model": "ling-3.0-flash",
      "name": "inclusionAI: Ling 3.0 Flash",
      "blurb": "*Ling-3.0-flash* is a *124B-parameter Mixture-of-Experts (MoE) model*, with approximately *5.1B parameters activated per token*. The model is designed with *token efficiency and production-scale agentic inference* as key priorities, enabling developers...",
      "contextWindow": 262144,
      "maxOutput": 32768,
      "cost": {
        "input": 0.042,
        "output": 0.126,
        "cacheRead": 0.0084
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-07-23",
      "openWeights": {
        "huggingFace": "inclusionAI/Ling-3.0-flash"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "inclusionai/ling-3.0-flash-vl",
      "orId": "inclusionai/ling-3.0-flash-vl",
      "vendor": "inclusionai",
      "model": "ling-3.0-flash-vl",
      "name": "inclusionAI: Ling 3.0 Flash VL",
      "blurb": "Ling 3.0 Flash VL builds on Ling 3.0 Flash (124B total / 5.5B active MoE from InclusionAI), further strengthening its language capabilities while adding native visual perception and advanced visual...",
      "contextWindow": 131072,
      "maxOutput": 32768,
      "cost": {
        "input": 0.12,
        "output": 0.36,
        "cacheRead": 0.024
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-09-10",
      "openWeights": {
        "huggingFace": "inclusionAI/Ling-3.0-flash-VL"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "inclusionai/ling-3.0-tiny",
      "orId": "inclusionai/ling-3.0-tiny",
      "vendor": "inclusionai",
      "model": "ling-3.0-tiny",
      "name": "inclusionAI: Ling 3.0 Tiny",
      "blurb": "Ling 3.0 Tiny is a mixture-of-experts model from InclusionAI, with 1.3B active parameters out of 7.9B total. It is designed for responsive agents, instruction following, and multi-turn conversations, with switchable...",
      "contextWindow": 262144,
      "maxOutput": 250000,
      "cost": {
        "input": 0.03,
        "output": 0.09,
        "cacheRead": 0.006
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-08-06",
      "openWeights": {
        "huggingFace": "inclusionAI/Ling-3.0-tiny"
      }
    },
    {
      "id": "inclusionai/ling-3.1-flash",
      "orId": "inclusionai/ling-3.1-flash",
      "vendor": "inclusionai",
      "model": "ling-3.1-flash",
      "name": "inclusionAI: Ling 3.1 Flash",
      "blurb": "Ling 3.1 Flash is a hybrid reasoning mixture-of-experts model from inclusionAI, with 25B active parameters out of 560B total.",
      "contextWindow": 1000000,
      "maxOutput": 262144,
      "cost": {
        "input": 0.09,
        "output": 0.27,
        "cacheRead": 0.018
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-10-02",
      "openWeights": {
        "huggingFace": "inclusionAI/Ling-3.1-flash"
      }
    },
    {
      "id": "inclusionai/ling-3.2-flash:free",
      "orId": "inclusionai/ling-3.2-flash:free",
      "vendor": "inclusionai",
      "model": "ling-3.2-flash:free",
      "name": "inclusionAI: Ling 3.2 Flash (free)",
      "blurb": "Ling 3.2 Flash is a hybrid reasoning model from inclusionAI that accepts text and image input, with a 1M-token context window and up to 200 images per request. Thinking can...",
      "contextWindow": 1048576,
      "maxOutput": 262144,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-11-13",
      "openWeights": {
        "huggingFace": "inclusionAI/Ling-3.2-flash"
      }
    },
    {
      "id": "inference-net/schematron-v2-small",
      "orId": "inference-net/schematron-v2-small",
      "vendor": "inference-net",
      "model": "schematron-v2-small",
      "name": "Inference.net: Schematron V2 Small",
      "blurb": "Schematron V2 Small is a 3B-parameter HTML-to-JSON extraction model from Inference.net. It prioritizes extraction quality for complex schemas and long pages. Extraction instructions must be supplied through a JSON schema...",
      "contextWindow": 128000,
      "maxOutput": 8192,
      "cost": {
        "input": 0.1,
        "output": 0.46,
        "cacheRead": 0.1
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2026-09-12",
      "openWeights": {
        "huggingFace": "inference-net/schematron-v2-llama-3.2-3b"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "inference-net/schematron-v2-turbo",
      "orId": "inference-net/schematron-v2-turbo",
      "vendor": "inference-net",
      "model": "schematron-v2-turbo",
      "name": "Inference.net: Schematron V2 Turbo",
      "blurb": "Schematron V2 Turbo is a 3B-parameter HTML-to-JSON extraction model from Inference.net. It prioritizes throughput for high-volume extraction workloads. Extraction instructions must be supplied through a JSON schema in response_format rather...",
      "contextWindow": 128000,
      "maxOutput": 8192,
      "cost": {
        "input": 0.06,
        "output": 0.3,
        "cacheRead": 0.06
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2026-09-12",
      "openWeights": {
        "huggingFace": "inference-net/schematron-v2-granite-4.0-h-micro"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "kt/midm-k-3.0-pro",
      "orId": "kt/midm-k-3.0-pro",
      "vendor": "kt",
      "model": "midm-k-3.0-pro",
      "name": "KT: Mi:dm K 3.0 Pro",
      "blurb": "Mi:dm K 3.0 Pro is a Mixture-of-Experts reasoning model from KT, with 120B total and 12B active parameters. It uses a hybrid architecture that mixes Mamba, MoE and attention layers.",
      "contextWindow": 261795,
      "maxOutput": 235615,
      "cost": {
        "input": 0.4,
        "output": 1.6,
        "cacheRead": 0.08
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-11-05"
    },
    {
      "id": "liquid/lfm-2.5-24b-a2b",
      "orId": "liquid/lfm-2.5-24b-a2b",
      "vendor": "liquid",
      "model": "lfm-2.5-24b-a2b",
      "name": "LiquidAI: LFM2.5-24B-A2B",
      "blurb": "LFM2.5-24B-A2B is a mixture-of-experts reasoning model from Liquid AI with 24B total and 2.45B active parameters. It always reasons before answering and is built for agents that call tools. Liquid...",
      "contextWindow": 128000,
      "maxOutput": 8192,
      "cost": {
        "input": 0.1,
        "output": 1.4,
        "cacheRead": 0.05
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-10-13",
      "openWeights": {
        "huggingFace": "LiquidAI/LFM2.5-24B-A2B"
      }
    },
    {
      "id": "liquid/lfm-3-1.7b",
      "orId": "liquid/lfm-3-1.7b",
      "vendor": "liquid",
      "model": "lfm-3-1.7b",
      "name": "LiquidAI: LFM3-1.7B",
      "blurb": "A compact 1.7B-parameter text model with a 256K-token context, post-trained for agents, tool calling, and structured outputs. It reasons before it answers. Prompts and outputs may be retained and used...",
      "contextWindow": 262144,
      "maxOutput": 8192,
      "cost": {
        "input": 0.16,
        "output": 0.4,
        "cacheRead": 0.04
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-11-06",
      "openWeights": {
        "huggingFace": "LiquidAI/LFM3-1.7B"
      }
    },
    {
      "id": "liquid/lfm-3-1.7b:free",
      "orId": "liquid/lfm-3-1.7b:free",
      "vendor": "liquid",
      "model": "lfm-3-1.7b:free",
      "name": "LiquidAI: LFM3-1.7B (free)",
      "blurb": "A compact 1.7B-parameter text model with a 256K-token context, post-trained for agents, tool calling, and structured outputs. It reasons before it answers. Prompts and outputs may be retained and used...",
      "contextWindow": 262144,
      "maxOutput": 8192,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-11-06",
      "openWeights": {
        "huggingFace": "LiquidAI/LFM3-1.7B"
      }
    },
    {
      "id": "mancer/weaver",
      "orId": "mancer/weaver",
      "vendor": "mancer",
      "model": "weaver",
      "name": "Mancer: Weaver (alpha)",
      "blurb": "An attempt to recreate Claude-style verbosity, but don't expect the same level of coherence or memory. Meant for use in roleplay/narrative situations.",
      "contextWindow": 8000,
      "maxOutput": 6000,
      "cost": {
        "input": 0.6,
        "output": 1,
        "cacheRead": 0.6
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2023-06-30",
      "released": "2023-08-02"
    },
    {
      "id": "meituan/longcat-2.0",
      "orId": "meituan/longcat-2.0",
      "vendor": "meituan",
      "model": "longcat-2.0",
      "name": "Meituan: LongCat 2.0",
      "blurb": "LongCat 2.0 is a sparse mixture-of-experts language model from Meituan, with 48B active parameters out of 1.6T total. It is suited for coding, repository-level changes, long-horizon problem solving, and agentic...",
      "contextWindow": 1048756,
      "maxOutput": 262144,
      "cost": {
        "input": 0.6,
        "output": 2.4,
        "cacheRead": 0.012
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-07-20",
      "openWeights": {
        "huggingFace": "meituan-longcat/LongCat-2.0"
      }
    },
    {
      "id": "meta-llama/llama-3.1-70b-instruct",
      "orId": "meta-llama/llama-3.1-70b-instruct",
      "vendor": "meta-llama",
      "model": "llama-3.1-70b-instruct",
      "name": "Meta: Llama 3.1 70B Instruct",
      "blurb": "Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors. This 70B instruct-tuned version is optimized for high quality dialogue usecases. It has demonstrated strong...",
      "contextWindow": 131072,
      "maxOutput": 8192,
      "cost": {
        "input": 1.44,
        "output": 1.44
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-12-31",
      "released": "2024-07-23",
      "openWeights": {
        "huggingFace": "meta-llama/Meta-Llama-3.1-70B-Instruct"
      },
      "bench": {
        "source": "epoch",
        "gpqa": 0.442,
        "gpqaStderr": 0.0245,
        "mathLevel5": 0.367,
        "mathLevel5Stderr": 0.0102,
        "aime": 0.036,
        "aimeStderr": 0.0169
      }
    },
    {
      "id": "meta-llama/llama-3.1-8b-instruct",
      "orId": "meta-llama/llama-3.1-8b-instruct",
      "vendor": "meta-llama",
      "model": "llama-3.1-8b-instruct",
      "name": "Meta: Llama 3.1 8B Instruct",
      "blurb": "Meta's latest class of model (Llama 3.1) launched with a variety of sizes & flavors. This 8B instruct-tuned version is fast and efficient. It has demonstrated strong performance compared to...",
      "contextWindow": 131072,
      "maxOutput": 16384,
      "cost": {
        "input": 0.04,
        "output": 0.08
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-12-31",
      "released": "2024-07-23",
      "openWeights": {
        "huggingFace": "meta-llama/Meta-Llama-3.1-8B-Instruct"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "gpqa": 0.27,
        "gpqaStderr": 0.0186,
        "mathLevel5": 0.229,
        "mathLevel5Stderr": 0.0085,
        "aime": 0.017,
        "aimeStderr": 0.0094
      }
    },
    {
      "id": "meta-llama/llama-3.2-1b-instruct",
      "orId": "meta-llama/llama-3.2-1b-instruct",
      "vendor": "meta-llama",
      "model": "llama-3.2-1b-instruct",
      "name": "Meta: Llama 3.2 1B Instruct",
      "blurb": "Llama 3.2 1B is a 1-billion-parameter language model focused on efficiently performing natural language tasks, such as summarization, dialogue, and multilingual text analysis. Its smaller size allows it to operate...",
      "contextWindow": 60000,
      "maxOutput": 54000,
      "cost": {
        "input": 0.054,
        "output": 0.402
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2023-12-31",
      "released": "2024-09-25",
      "retiresAt": "2026-11-20",
      "openWeights": {
        "huggingFace": "meta-llama/Llama-3.2-1B-Instruct"
      },
      "bench": {
        "source": "epoch",
        "gpqa": 0.239,
        "gpqaStderr": 0.0163,
        "aime": 0.006,
        "aimeStderr": 0.0039
      }
    },
    {
      "id": "meta-llama/llama-3.2-3b-instruct",
      "orId": "meta-llama/llama-3.2-3b-instruct",
      "vendor": "meta-llama",
      "model": "llama-3.2-3b-instruct",
      "name": "Meta: Llama 3.2 3B Instruct",
      "blurb": "Llama 3.2 3B is a 3-billion-parameter multilingual large language model, optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization. Designed with the latest transformer architecture, it...",
      "contextWindow": 131072,
      "maxOutput": 72000,
      "cost": {
        "input": 0.1018,
        "output": 0.67
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2023-12-31",
      "released": "2024-09-25",
      "retiresAt": "2026-11-20",
      "openWeights": {
        "huggingFace": "meta-llama/Llama-3.2-3B-Instruct"
      }
    },
    {
      "id": "meta-llama/llama-3.3-70b-instruct",
      "orId": "meta-llama/llama-3.3-70b-instruct",
      "vendor": "meta-llama",
      "model": "llama-3.3-70b-instruct",
      "name": "Meta: Llama 3.3 70B Instruct",
      "blurb": "The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out). The Llama 3.3 instruction tuned text only model...",
      "contextWindow": 131072,
      "maxOutput": 115200,
      "cost": {
        "input": 1.42,
        "output": 1.42,
        "cacheRead": 1.42
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-12-31",
      "released": "2024-12-06",
      "openWeights": {
        "huggingFace": "meta-llama/Llama-3.3-70B-Instruct"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "gpqa": 0.474,
        "gpqaStderr": 0.0285,
        "mathLevel5": 0.416,
        "mathLevel5Stderr": 0.0118,
        "aime": 0.051,
        "aimeStderr": 0.0236
      }
    },
    {
      "id": "meta-llama/llama-4-maverick",
      "orId": "meta-llama/llama-4-maverick",
      "vendor": "meta-llama",
      "model": "llama-4-maverick",
      "name": "Meta: Llama 4 Maverick",
      "blurb": "Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...",
      "contextWindow": 524288,
      "maxOutput": 32768,
      "cost": {
        "input": 0.7,
        "output": 2,
        "cacheRead": 0.34
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-08-31",
      "released": "2025-04-05",
      "openWeights": {
        "huggingFace": "meta-llama/Llama-4-Maverick-17B-128E-Instruct"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 132.08,
        "accessibility": "Open weights (restricted use)"
      }
    },
    {
      "id": "meta-llama/llama-4-scout",
      "orId": "meta-llama/llama-4-scout",
      "vendor": "meta-llama",
      "model": "llama-4-scout",
      "name": "Meta: Llama 4 Scout",
      "blurb": "Llama 4 Scout 17B Instruct (16E) is a mixture-of-experts (MoE) language model developed by Meta, activating 17 billion parameters out of a total of 109B. It supports native multimodal input...",
      "contextWindow": 1310720,
      "maxOutput": 16384,
      "cost": {
        "input": 0.2,
        "output": 0.6
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-08-31",
      "released": "2025-04-05",
      "openWeights": {
        "huggingFace": "meta-llama/Llama-4-Scout-17B-16E-Instruct"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 127.68,
        "accessibility": "Open weights (restricted use)"
      }
    },
    {
      "id": "meta-llama/llama-guard-4-12b",
      "orId": "meta-llama/llama-guard-4-12b",
      "vendor": "meta-llama",
      "model": "llama-guard-4-12b",
      "name": "Meta: Llama Guard 4 12B",
      "blurb": "Llama Guard 4 is a Llama 4 Scout-derived multimodal pretrained model, fine-tuned for content safety classification. Similar to previous versions, it can be used to classify content in both LLM...",
      "contextWindow": 163840,
      "maxOutput": 16384,
      "cost": {
        "input": 0.36,
        "output": 0.36
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-08-31",
      "released": "2025-04-30",
      "openWeights": {
        "huggingFace": "meta-llama/Llama-Guard-4-12B"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "meta/muse-ember",
      "orId": "meta/muse-ember",
      "vendor": "meta",
      "model": "muse-ember",
      "name": "Meta: Muse Ember",
      "blurb": "Muse Ember is the smallest and most efficient model from Meta that’s designed for complex reasoning execution and the work agents repeat all day: tool calls, coding agent subtasks, routing,...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.2,
        "output": 1,
        "cacheRead": 0.008,
        "cacheWrite": 0.25,
        "webSearch": 0.005
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-11-09"
    },
    {
      "id": "meta/muse-glimmer-30b",
      "orId": "meta/muse-glimmer-30b",
      "vendor": "meta",
      "model": "muse-glimmer-30b",
      "name": "Meta: Muse Glimmer 30B",
      "blurb": "Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware. It is suited for long-horizon...",
      "contextWindow": 131072,
      "maxOutput": 16384,
      "cost": {
        "input": 0.6,
        "output": 2.4,
        "cacheRead": 0.08
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-08-09",
      "openWeights": {
        "huggingFace": "meta-models/Muse-Glimmer-30B"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "meta/muse-glimmer-30b:batch",
      "orId": "meta/muse-glimmer-30b:batch",
      "vendor": "meta",
      "model": "muse-glimmer-30b:batch",
      "name": "Meta: Muse Glimmer 30B (batch)",
      "blurb": "Muse Glimmer 30B is a dense, open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark and optimized for autonomous agents on consumer hardware. It is suited for long-horizon...",
      "contextWindow": 131072,
      "maxOutput": 16384,
      "cost": {
        "input": 0.48,
        "output": 1.92,
        "cacheRead": 0.064
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-08-09",
      "openWeights": {
        "huggingFace": "meta-models/Muse-Glimmer-30B"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "meta/muse-spark-1.1",
      "orId": "meta/muse-spark-1.1",
      "vendor": "meta",
      "model": "muse-spark-1.1",
      "name": "Meta: Muse Spark 1.1",
      "blurb": "Muse Spark 1.1 is a multimodal reasoning model from Meta, built for agentic tasks. It accepts text, images, video, and PDF documents and returns text, with a 1M-token context window....",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 2.5,
        "output": 8.5,
        "cacheRead": 0.3,
        "webSearch": 0.005
      },
      "input": [
        "text",
        "image",
        "video",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-07-16",
      "bench": {
        "source": "epoch",
        "eci": 154.87,
        "accessibility": "API access",
        "simpleQA": 0.578,
        "simpleQAStderr": 0.0157
      }
    },
    {
      "id": "meta/muse-spark-1.2",
      "orId": "meta/muse-spark-1.2",
      "vendor": "meta",
      "model": "muse-spark-1.2",
      "name": "Meta: Muse Spark 1.2",
      "blurb": "Muse Spark 1.2 is a reasoning model from Meta, designed for complex agentic tasks. It accepts text, images, video, and PDF documents, returns text, and offers a 1M-token context window....",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 2.5,
        "output": 8.5,
        "cacheRead": 0.3,
        "webSearch": 0.005
      },
      "input": [
        "text",
        "image",
        "video",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-08-05",
      "bench": {
        "source": "epoch",
        "eci": 155.04,
        "accessibility": "API access",
        "simpleQA": 0.603,
        "simpleQAStderr": 0.0155
      }
    },
    {
      "id": "meta/muse-spark-1.2-contributor",
      "orId": "meta/muse-spark-1.2-contributor",
      "vendor": "meta",
      "model": "muse-spark-1.2-contributor",
      "name": "Meta: Muse Spark 1.2 Contributor",
      "blurb": "Muse Spark 1.2 contributor tier is a reasoning model from Meta designed for developers who want to start building at an even lower cost. It’s meaningfully cheaper than Muse Spark...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.2,
        "output": 0.4,
        "cacheRead": 0.004,
        "webSearch": 0.005
      },
      "input": [
        "text",
        "image",
        "video",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-08-21"
    },
    {
      "id": "meta/muse-spark-1.3",
      "orId": "meta/muse-spark-1.3",
      "vendor": "meta",
      "model": "muse-spark-1.3",
      "name": "Meta: Muse Spark 1.3",
      "blurb": "Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It is designed to keep track of information across extended tasks, work through...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 2.5,
        "output": 8.5,
        "cacheRead": 0.3,
        "webSearch": 0.005
      },
      "input": [
        "text",
        "image",
        "video",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-02",
      "bench": {
        "source": "epoch",
        "eci": 158.24,
        "accessibility": "API access",
        "aime": 0.992,
        "aimeStderr": 0.0062
      }
    },
    {
      "id": "meta/muse-spark-1.3-contributor",
      "orId": "meta/muse-spark-1.3-contributor",
      "vendor": "meta",
      "model": "muse-spark-1.3-contributor",
      "name": "Meta: Muse Spark 1.3 Contributor",
      "blurb": "Muse Spark 1.3 Contributor is the cost-efficient contributor tier of Meta’s multimodal reasoning model for experimentation, learning, and early-stage agentic, multi-agent, and coding workflows. It is designed to track information...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.2,
        "output": 0.4,
        "cacheRead": 0.004,
        "webSearch": 0.005
      },
      "input": [
        "text",
        "image",
        "video",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-02"
    },
    {
      "id": "meta/muse-spark-open",
      "orId": "meta/muse-spark-open",
      "vendor": "meta",
      "model": "muse-spark-open",
      "name": "Meta: Muse Spark Open",
      "blurb": "Muse Spark Open is a coding-optimized open-weight model. Trained specifically for agentic coding workflows, it delivers higher first-attempt accuracy, reliable tool calling, and deep codebase understanding across the repos developers...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 2.5,
        "output": 8.5,
        "cacheRead": 0.3
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-15",
      "openWeights": {
        "huggingFace": "meta-models/Muse-Spark-Open-816B-A42B"
      },
      "bench": {
        "source": "epoch",
        "eci": 155.42,
        "accessibility": "Open weights (restricted use)"
      }
    },
    {
      "id": "meta/muse-wonder",
      "orId": "meta/muse-wonder",
      "vendor": "meta",
      "model": "muse-wonder",
      "name": "Meta: Muse Wonder",
      "blurb": "Muse Wonder is a frontier reasoning model from Meta, designed for complex personal and professional agentic tasks. It accepts text, images, video, audio, and PDF documents and returns text, with...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 8,
        "output": 40,
        "cacheRead": 0.4,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "video",
        "file",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-12",
      "bench": {
        "source": "epoch",
        "eci": 161.68,
        "accessibility": "API access"
      }
    },
    {
      "id": "microsoft/phi-4",
      "orId": "microsoft/phi-4",
      "vendor": "microsoft",
      "model": "phi-4",
      "name": "Microsoft: Phi 4",
      "blurb": "[Microsoft Research](/microsoft) Phi-4 is designed to perform well in complex reasoning tasks and can operate efficiently in situations with limited memory or where quick responses are needed. At 14 billion...",
      "contextWindow": 16384,
      "maxOutput": 14745,
      "cost": {
        "input": 0.14,
        "output": 0.28
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-01-10",
      "openWeights": {
        "huggingFace": "microsoft/phi-4"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 129.04,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.561,
        "gpqaStderr": 0.0259,
        "mathLevel5": 0.649,
        "mathLevel5Stderr": 0.0107,
        "aime": 0.138,
        "aimeStderr": 0.037
      }
    },
    {
      "id": "minimax/minimax-m2.5",
      "orId": "minimax/minimax-m2.5",
      "vendor": "minimax",
      "model": "minimax-m2.5",
      "name": "MiniMax: MiniMax M2.5",
      "blurb": "MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digital working environments, M2.5 builds upon the coding expertise of M2.1...",
      "contextWindow": 198000,
      "maxOutput": 176947,
      "cost": {
        "input": 0.59,
        "output": 2.4,
        "cacheRead": 0.12
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-02-12",
      "openWeights": {
        "huggingFace": "MiniMaxAI/MiniMax-M2.5"
      }
    },
    {
      "id": "minimax/minimax-m2.7",
      "orId": "minimax/minimax-m2.7",
      "vendor": "minimax",
      "model": "minimax-m2.7",
      "name": "MiniMax: MiniMax M2.7",
      "blurb": "MiniMax-M2.7 is a next-generation large language model designed for autonomous, real-world productivity and continuous improvement. Built to actively participate in its own evolution, M2.7 integrates advanced agentic capabilities through multi-agent...",
      "contextWindow": 196608,
      "maxOutput": 176947,
      "cost": {
        "input": 0.42,
        "output": 1.68,
        "cacheRead": 0.084
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-03-18",
      "openWeights": {
        "huggingFace": "MiniMaxAI/MiniMax-M2.7"
      }
    },
    {
      "id": "minimax/minimax-m3",
      "orId": "minimax/minimax-m3",
      "vendor": "minimax",
      "model": "minimax-m3",
      "name": "MiniMax: MiniMax M3",
      "blurb": "MiniMax-M3 is a multimodal foundation model from MiniMax. It supports text, image, and video inputs with text output, a 1M-token context window, and is suited for long-horizon agentic work, coding,...",
      "contextWindow": 1048576,
      "maxOutput": 512000,
      "cost": {
        "input": 0.6,
        "output": 2.4,
        "cacheRead": 0.12
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-05-31",
      "openWeights": {
        "huggingFace": "MiniMaxAI/Minimax-M3"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 147.74,
        "accessibility": "Open weights (restricted use)",
        "gpqa": 0.909,
        "gpqaStderr": 0.0205,
        "aime": 0.711,
        "aimeStderr": 0.0683
      }
    },
    {
      "id": "minimax/minimax-m3:batch",
      "orId": "minimax/minimax-m3:batch",
      "vendor": "minimax",
      "model": "minimax-m3:batch",
      "name": "MiniMax: MiniMax M3 (batch)",
      "blurb": "MiniMax-M3 is a multimodal foundation model from MiniMax. It supports text, image, and video inputs with text output, a 1M-token context window, and is suited for long-horizon agentic work, coding,...",
      "contextWindow": 524288,
      "maxOutput": 512000,
      "cost": {
        "input": 0.448,
        "output": 1.76,
        "cacheRead": 0.0896
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-05-31",
      "openWeights": {
        "huggingFace": "MiniMaxAI/Minimax-M3"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "minimax/minimax-m3.1-flash",
      "orId": "minimax/minimax-m3.1-flash",
      "vendor": "minimax",
      "model": "minimax-m3.1-flash",
      "name": "MiniMax: MiniMax M3.1 Flash",
      "blurb": "MiniMax-M3.1-Flash is a multimodal model from MiniMax. Compared with M3, it delivers significant improvements across multiple benchmarks, particularly in complex agentic tasks, coding, and cybersecurity. It is built for complex,...",
      "contextWindow": 1048576,
      "maxOutput": 524288,
      "cost": {
        "input": 0.3,
        "output": 1.2,
        "cacheRead": 0.03,
        "tiers": [
          {
            "input": 0.15,
            "output": 0.6,
            "cacheRead": 0.015
          },
          {
            "input": 0.3,
            "output": 1.2,
            "cacheRead": 0.03
          }
        ]
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-10-19",
      "openWeights": {
        "huggingFace": "MiniMaxAI/MiniMax-M3.1-Flash"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "mistralai/codestral-2508",
      "orId": "mistralai/codestral-2508",
      "vendor": "mistralai",
      "model": "codestral-2508",
      "name": "Mistral: Codestral 2508",
      "blurb": "Mistral's cutting-edge language model for coding released end of July 2025. Codestral specializes in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code correction and test generation.\n\n[Blog Post](https://mistral.ai/news/codestral-25-08)",
      "contextWindow": 256000,
      "maxOutput": 204800,
      "cost": {
        "input": 0.6,
        "output": 1.8,
        "cacheRead": 0.06
      },
      "input": [
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-08-01",
      "regions": [
        "eu"
      ]
    },
    {
      "id": "mistralai/devstral-2512",
      "orId": "mistralai/devstral-2512",
      "vendor": "mistralai",
      "model": "devstral-2512",
      "name": "Mistral: Devstral 2 2512",
      "blurb": "Devstral 2 is a state-of-the-art open-source model by Mistral AI specializing in agentic coding. It is a 123B-parameter dense transformer model supporting a 256K context window. Devstral 2 supports exploring...",
      "contextWindow": 262144,
      "maxOutput": 209715,
      "cost": {
        "input": 0.8,
        "output": 4,
        "cacheRead": 0.08
      },
      "input": [
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2025-12-09",
      "openWeights": {
        "huggingFace": "mistralai/Devstral-2-123B-Instruct-2512"
      },
      "regions": [
        "eu"
      ]
    },
    {
      "id": "mistralai/ministral-14b-2512",
      "orId": "mistralai/ministral-14b-2512",
      "vendor": "mistralai",
      "model": "ministral-14b-2512",
      "name": "Mistral: Ministral 3 14B 2512",
      "blurb": "The largest model in the Ministral 3 family, Ministral 3 14B offers frontier capabilities and performance comparable to its larger Mistral Small 3.2 24B counterpart. A powerful and efficient language...",
      "contextWindow": 262144,
      "maxOutput": 209715,
      "cost": {
        "input": 0.4,
        "output": 0.4,
        "cacheRead": 0.04
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2025-12-02",
      "openWeights": {
        "huggingFace": "mistralai/Ministral-3-14B-Instruct-2512"
      },
      "regions": [
        "eu"
      ]
    },
    {
      "id": "mistralai/ministral-3b-2512",
      "orId": "mistralai/ministral-3b-2512",
      "vendor": "mistralai",
      "model": "ministral-3b-2512",
      "name": "Mistral: Ministral 3 3B 2512",
      "blurb": "The smallest model in the Ministral 3 family, Ministral 3 3B is a powerful, efficient tiny language model with vision capabilities.",
      "contextWindow": 131072,
      "maxOutput": 104857,
      "cost": {
        "input": 0.2,
        "output": 0.2,
        "cacheRead": 0.02
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2025-12-02",
      "openWeights": {
        "huggingFace": "mistralai/Ministral-3-3B-Instruct-2512"
      },
      "regions": [
        "eu"
      ]
    },
    {
      "id": "mistralai/ministral-8b-2512",
      "orId": "mistralai/ministral-8b-2512",
      "vendor": "mistralai",
      "model": "ministral-8b-2512",
      "name": "Mistral: Ministral 3 8B 2512",
      "blurb": "A balanced model in the Ministral 3 family, Ministral 3 8B is a powerful, efficient tiny language model with vision capabilities.",
      "contextWindow": 262144,
      "maxOutput": 209715,
      "cost": {
        "input": 0.3,
        "output": 0.3,
        "cacheRead": 0.03
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2025-12-02",
      "openWeights": {
        "huggingFace": "mistralai/Ministral-3-8B-Instruct-2512"
      },
      "regions": [
        "eu"
      ]
    },
    {
      "id": "mistralai/mistral-large",
      "orId": "mistralai/mistral-large",
      "vendor": "mistralai",
      "model": "mistral-large",
      "name": "Mistral Large",
      "blurb": "This is Mistral AI's flagship model, Mistral Large 2 (version `mistral-large-2407`). It's a proprietary weights-available model and excels at reasoning, code, JSON, chat, and more. Read the launch announcement [here](https://mistral.ai/news/mistral-large-2407/)....",
      "contextWindow": 128000,
      "maxOutput": 102400,
      "cost": {
        "input": 4,
        "output": 12,
        "cacheRead": 0.4
      },
      "input": [
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-11-30",
      "released": "2024-02-26",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 119.81,
        "accessibility": "API access"
      }
    },
    {
      "id": "mistralai/mistral-large-2407",
      "orId": "mistralai/mistral-large-2407",
      "vendor": "mistralai",
      "model": "mistral-large-2407",
      "name": "Mistral Large 2407",
      "blurb": "This is Mistral AI's flagship model, Mistral Large 2 (version mistral-large-2407). It's a proprietary weights-available model and excels at reasoning, code, JSON, chat, and more. Read the launch announcement [here](https://mistral.ai/news/mistral-large-2407/)....",
      "contextWindow": 131072,
      "maxOutput": 104857,
      "cost": {
        "input": 4,
        "output": 12,
        "cacheRead": 0.4
      },
      "input": [
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-03-31",
      "released": "2024-11-19",
      "openWeights": {
        "huggingFace": "mistralai/Mistral-Large-Instruct-2407"
      },
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "gpqa": 0.49,
        "gpqaStderr": 0.0259,
        "mathLevel5": 0.448,
        "mathLevel5Stderr": 0.0107,
        "aime": 0.085,
        "aimeStderr": 0.0247
      }
    },
    {
      "id": "mistralai/mistral-large-2512",
      "orId": "mistralai/mistral-large-2512",
      "vendor": "mistralai",
      "model": "mistral-large-2512",
      "name": "Mistral: Mistral Large 3 2512",
      "blurb": "Mistral Large 3 2512 is Mistral’s most capable model to date, featuring a sparse mixture-of-experts architecture with 41B active parameters (675B total), and released under the Apache 2.0 license.",
      "contextWindow": 262144,
      "maxOutput": 209715,
      "cost": {
        "input": 1,
        "output": 3,
        "cacheRead": 0.1
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2025-12-01",
      "openWeights": {
        "huggingFace": "mistralai/Mistral-Large-3-675B-Instruct-2512"
      },
      "regions": [
        "eu"
      ]
    },
    {
      "id": "mistralai/mistral-large-4-0",
      "orId": "mistralai/mistral-large-4-0",
      "vendor": "mistralai",
      "model": "mistral-large-4-0",
      "name": "Mistral: Mistral Large 4",
      "blurb": "Mistral Large 4 is a frontier multimodal (text and image input) model from Mistral AI built for reasoning, coding, and agentic workloads. It offers a 1M-token context window with up...",
      "contextWindow": 1048576,
      "maxOutput": 262144,
      "cost": {
        "input": 1.4,
        "output": 4.2,
        "cacheRead": 0.14
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-10-06",
      "openWeights": {
        "huggingFace": "mistralai/Mistral-Large-4-1T-A52B"
      },
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 154.62,
        "accessibility": "Open weights (restricted use)",
        "gpqa": 0.881,
        "gpqaStderr": 0.0193,
        "aime": 0.983,
        "aimeStderr": 0.0064,
        "simpleQA": 0.2,
        "simpleQAStderr": 0.0127
      }
    },
    {
      "id": "mistralai/mistral-medium-3",
      "orId": "mistralai/mistral-medium-3",
      "vendor": "mistralai",
      "model": "mistral-medium-3",
      "name": "Mistral: Mistral Medium 3",
      "blurb": "Mistral Medium 3 is a high-performance enterprise-grade language model designed to deliver frontier-level capabilities at significantly reduced operational cost. It balances state-of-the-art reasoning and multimodal performance with 8× lower cost...",
      "contextWindow": 131072,
      "maxOutput": 104857,
      "cost": {
        "input": 0.8,
        "output": 4,
        "cacheRead": 0.08
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-05-07",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 132.81,
        "accessibility": "API access"
      }
    },
    {
      "id": "mistralai/mistral-medium-3-5",
      "orId": "mistralai/mistral-medium-3-5",
      "vendor": "mistralai",
      "model": "mistral-medium-3-5",
      "name": "Mistral: Mistral Medium 3.5",
      "blurb": "Mistral Medium 3.5 is a dense 128B instruction-following model from Mistral AI. It supports text and image inputs with text output, and is designed for agentic workflows, coding, and complex...",
      "contextWindow": 262144,
      "maxOutput": 209715,
      "cost": {
        "input": 3,
        "output": 15
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-04-30",
      "openWeights": {
        "huggingFace": "mistralai/Mistral-Medium-3.5-128B"
      },
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 141.59,
        "accessibility": "Open weights (restricted use)"
      }
    },
    {
      "id": "mistralai/mistral-medium-3-5:batch",
      "orId": "mistralai/mistral-medium-3-5:batch",
      "vendor": "mistralai",
      "model": "mistral-medium-3-5:batch",
      "name": "Mistral: Mistral Medium 3.5 (batch)",
      "blurb": "Mistral Medium 3.5 is a dense 128B instruction-following model from Mistral AI. It supports text and image inputs with text output, and is designed for agentic workflows, coding, and complex...",
      "contextWindow": 262144,
      "maxOutput": 209715,
      "cost": {
        "input": 1.5,
        "output": 7.5
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-04-30",
      "openWeights": {
        "huggingFace": "mistralai/Mistral-Medium-3.5-128B"
      }
    },
    {
      "id": "mistralai/mistral-medium-3.1",
      "orId": "mistralai/mistral-medium-3.1",
      "vendor": "mistralai",
      "model": "mistral-medium-3.1",
      "name": "Mistral: Mistral Medium 3.1",
      "blurb": "Mistral Medium 3.1 is an updated version of Mistral Medium 3, which is a high-performance enterprise-grade language model designed to deliver frontier-level capabilities at significantly reduced operational cost. It balances...",
      "contextWindow": 131072,
      "maxOutput": 104857,
      "cost": {
        "input": 0.8,
        "output": 4,
        "cacheRead": 0.08
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-06-30",
      "released": "2025-08-13",
      "regions": [
        "eu"
      ]
    },
    {
      "id": "mistralai/mistral-nemo",
      "orId": "mistralai/mistral-nemo",
      "vendor": "mistralai",
      "model": "mistral-nemo",
      "name": "Mistral: Mistral Nemo",
      "blurb": "A 12B parameter model with a 128k token context length built by Mistral in collaboration with NVIDIA. The model is multilingual, supporting English, French, German, Spanish, Italian, Portuguese, Chinese, Japanese,...",
      "contextWindow": 131072,
      "maxOutput": 16384,
      "cost": {
        "input": 0.058,
        "output": 0.06
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-04-30",
      "released": "2024-07-19",
      "openWeights": {
        "huggingFace": "mistralai/Mistral-Nemo-Instruct-2407"
      },
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 113.88,
        "accessibility": "Open weights (unrestricted)"
      }
    },
    {
      "id": "mistralai/mistral-saba",
      "orId": "mistralai/mistral-saba",
      "vendor": "mistralai",
      "model": "mistral-saba",
      "name": "Mistral: Saba",
      "blurb": "Mistral Saba is a 24B-parameter language model specifically designed for the Middle East and South Asia, delivering accurate and contextually relevant responses while maintaining efficient performance. Trained on curated regional...",
      "contextWindow": 32768,
      "maxOutput": 26214,
      "cost": {
        "input": 0.4,
        "output": 1.2,
        "cacheRead": 0.04
      },
      "input": [
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-09-30",
      "released": "2025-02-17",
      "regions": [
        "eu"
      ]
    },
    {
      "id": "mistralai/mistral-small-24b-instruct-2501",
      "orId": "mistralai/mistral-small-24b-instruct-2501",
      "vendor": "mistralai",
      "model": "mistral-small-24b-instruct-2501",
      "name": "Mistral: Mistral Small 3",
      "blurb": "Mistral Small 3 is a 24B-parameter language model optimized for low-latency performance across common AI tasks. Released under the Apache 2.0 license, it features both pre-trained and instruction-tuned versions designed...",
      "contextWindow": 32768,
      "maxOutput": 16384,
      "cost": {
        "input": 0.1,
        "output": 0.16
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2023-10-31",
      "released": "2025-01-30",
      "openWeights": {
        "huggingFace": "mistralai/Mistral-Small-24B-Instruct-2501"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "mistralai/mistral-small-2603",
      "orId": "mistralai/mistral-small-2603",
      "vendor": "mistralai",
      "model": "mistral-small-2603",
      "name": "Mistral: Mistral Small 4",
      "blurb": "Mistral Small 4 is the next major release in the Mistral Small family, unifying the capabilities of several flagship Mistral models into a single system. It combines strong reasoning from...",
      "contextWindow": 262144,
      "maxOutput": 209715,
      "cost": {
        "input": 0.3,
        "output": 1.2,
        "cacheRead": 0.03
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-03-16",
      "openWeights": {
        "huggingFace": "mistralai/Mistral-Small-4-119B-2603"
      },
      "regions": [
        "eu",
        "us"
      ]
    },
    {
      "id": "mistralai/mistral-small-3.1-24b-instruct",
      "orId": "mistralai/mistral-small-3.1-24b-instruct",
      "vendor": "mistralai",
      "model": "mistral-small-3.1-24b-instruct",
      "name": "Mistral: Mistral Small 3.1 24B",
      "blurb": "Mistral Small 3.1 24B Instruct is an upgraded variant of Mistral Small 3 (2501), featuring 24 billion parameters with advanced multimodal capabilities. It provides state-of-the-art performance in text-based reasoning and...",
      "contextWindow": 128000,
      "maxOutput": 102400,
      "cost": {
        "input": 0.702,
        "output": 1.11
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2025-03-17",
      "openWeights": {
        "huggingFace": "mistralai/Mistral-Small-3.1-24B-Instruct-2503"
      }
    },
    {
      "id": "mistralai/mistral-small-3.2-24b-instruct",
      "orId": "mistralai/mistral-small-3.2-24b-instruct",
      "vendor": "mistralai",
      "model": "mistral-small-3.2-24b-instruct",
      "name": "Mistral: Mistral Small 3.2 24B",
      "blurb": "Mistral-Small-3.2-24B-Instruct-2506 is an updated 24B parameter model from Mistral optimized for instruction following, repetition reduction, and improved function calling. Compared to the 3.1 release, version 3.2 significantly improves accuracy on...",
      "contextWindow": 256000,
      "maxOutput": 26214,
      "cost": {
        "input": 0.2,
        "output": 0.6,
        "cacheRead": 0.02
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2025-06-20",
      "openWeights": {
        "huggingFace": "mistralai/Mistral-Small-3.2-24B-Instruct-2506"
      },
      "regions": [
        "eu",
        "us"
      ]
    },
    {
      "id": "mistralai/mixtral-8x22b-instruct",
      "orId": "mistralai/mixtral-8x22b-instruct",
      "vendor": "mistralai",
      "model": "mixtral-8x22b-instruct",
      "name": "Mistral: Mixtral 8x22B Instruct",
      "blurb": "Mistral's official instruct fine-tuned version of [Mixtral 8x22B](/models/mistralai/mixtral-8x22b). It uses 39B active parameters out of 141B, offering unparalleled cost efficiency for its size. Its strengths include: - strong math, coding,...",
      "contextWindow": 65536,
      "maxOutput": 52428,
      "cost": {
        "input": 4,
        "output": 12,
        "cacheRead": 0.4
      },
      "input": [
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-01-31",
      "released": "2024-04-17",
      "openWeights": {
        "huggingFace": "mistralai/Mixtral-8x22B-Instruct-v0.1"
      },
      "regions": [
        "eu"
      ]
    },
    {
      "id": "mistralai/voxtral-small-24b-2507",
      "orId": "mistralai/voxtral-small-24b-2507",
      "vendor": "mistralai",
      "model": "voxtral-small-24b-2507",
      "name": "Mistral: Voxtral Small 24B 2507",
      "blurb": "Voxtral Small is an enhancement of Mistral Small 3, incorporating state-of-the-art audio input capabilities while retaining best-in-class text performance. It excels at speech transcription, translation and audio understanding. Input audio...",
      "contextWindow": 32768,
      "maxOutput": 26214,
      "cost": {
        "input": 0.2,
        "output": 0.6,
        "cacheRead": 0.02,
        "audioIn": 200
      },
      "input": [
        "text",
        "audio",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2025-10-30",
      "openWeights": {
        "huggingFace": "mistralai/Voxtral-Small-24B-2507"
      },
      "regions": [
        "eu"
      ]
    },
    {
      "id": "moondream/moondream3-preview",
      "orId": "moondream/moondream3-preview",
      "vendor": "moondream",
      "model": "moondream3-preview",
      "name": "Moondream: Moondream 3 Preview",
      "blurb": "Moondream 3 Preview is a vision-language model from Moondream with 9B total parameters and 2B active (mixture-of-experts). It handles visual question answering, captioning, OCR, and visual reasoning over images, with...",
      "contextWindow": 32000,
      "maxOutput": 4096,
      "cost": {
        "input": 0.6,
        "output": 5
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": false,
      "released": "2026-10-15",
      "openWeights": {
        "huggingFace": "moondream/moondream3-preview"
      }
    },
    {
      "id": "moondream/moondream3.1-9b-a2b",
      "orId": "moondream/moondream3.1-9b-a2b",
      "vendor": "moondream",
      "model": "moondream3.1-9b-a2b",
      "name": "Moondream: Moondream 3.1",
      "blurb": "Moondream 3.1 is a vision-language model from Moondream, the successor to Moondream 3 Preview, with 9B total parameters and 2B active (mixture-of-experts). It handles visual question answering, captioning, OCR, and...",
      "contextWindow": 32000,
      "maxOutput": 4096,
      "cost": {
        "input": 0.6,
        "output": 5
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": false,
      "released": "2026-10-15",
      "openWeights": {
        "huggingFace": "moondream/moondream3.1-9B-A2B"
      }
    },
    {
      "id": "moonshotai/kimi-k2",
      "orId": "moonshotai/kimi-k2",
      "vendor": "moonshotai",
      "model": "kimi-k2",
      "name": "MoonshotAI: Kimi K2 0711",
      "blurb": "Kimi K2 Instruct is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32 billion active per forward pass. It is optimized for...",
      "contextWindow": 131072,
      "maxOutput": 98304,
      "cost": {
        "input": 1.14,
        "output": 4.6
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-12-31",
      "released": "2025-07-11",
      "openWeights": {
        "huggingFace": "moonshotai/Kimi-K2-Instruct"
      }
    },
    {
      "id": "moonshotai/kimi-k2-0905",
      "orId": "moonshotai/kimi-k2-0905",
      "vendor": "moonshotai",
      "model": "kimi-k2-0905",
      "name": "MoonshotAI: Kimi K2 0905",
      "blurb": "Kimi K2 0905 is the September update of [Kimi K2 0711](moonshotai/kimi-k2). It is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32...",
      "contextWindow": 262144,
      "maxOutput": 98304,
      "cost": {
        "input": 1.2,
        "output": 5
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-12-31",
      "released": "2025-09-04",
      "openWeights": {
        "huggingFace": "moonshotai/Kimi-K2-Instruct-0905"
      }
    },
    {
      "id": "moonshotai/kimi-k2-thinking",
      "orId": "moonshotai/kimi-k2-thinking",
      "vendor": "moonshotai",
      "model": "kimi-k2-thinking",
      "name": "MoonshotAI: Kimi K2 Thinking",
      "blurb": "Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning. Built on the trillion-parameter Mixture-of-Experts (MoE) architecture introduced in...",
      "contextWindow": 262144,
      "maxOutput": 98304,
      "cost": {
        "input": 1.2,
        "output": 5,
        "cacheRead": 0.3
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2025-11-06",
      "openWeights": {
        "huggingFace": "moonshotai/Kimi-K2-Thinking"
      },
      "bench": {
        "source": "epoch",
        "eci": 145.43,
        "accessibility": "Open weights (restricted use)"
      }
    },
    {
      "id": "moonshotai/kimi-k2.5",
      "orId": "moonshotai/kimi-k2.5",
      "vendor": "moonshotai",
      "model": "kimi-k2.5",
      "name": "MoonshotAI: Kimi K2.5",
      "blurb": "Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm paradigm. Built on Kimi K2 with continued pretraining over approximately 15T mixed...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 1,
        "output": 5,
        "cacheRead": 0.3
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-01-27",
      "openWeights": {
        "huggingFace": "moonshotai/Kimi-K2.5"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 148.76,
        "accessibility": "Open weights (unrestricted)",
        "sweBenchVerified": 0.738,
        "sweBenchVerifiedStderr": 0.02,
        "simpleQA": 0.343,
        "simpleQAStderr": 0.015
      }
    },
    {
      "id": "moonshotai/kimi-k2.6",
      "orId": "moonshotai/kimi-k2.6",
      "vendor": "moonshotai",
      "model": "kimi-k2.6",
      "name": "MoonshotAI: Kimi K2.6",
      "blurb": "Kimi K2.6 is Moonshot AI's next-generation multimodal model, designed for long-horizon coding, coding-driven UI/UX generation, and multi-agent orchestration. It handles complex end-to-end coding tasks across Python, Rust, and Go, and...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 1.3,
        "output": 6.82,
        "cacheRead": 0.3
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-20",
      "openWeights": {
        "huggingFace": "moonshotai/Kimi-K2.6"
      },
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 151.56,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.908,
        "gpqaStderr": 0.0172,
        "sweBenchVerified": 0.767,
        "sweBenchVerifiedStderr": 0.0192,
        "frontierMath": 0.39,
        "frontierMathStderr": 0.0287,
        "aime": 0.961,
        "aimeStderr": 0.0238,
        "simpleQA": 0.349,
        "simpleQAStderr": 0.0151
      }
    },
    {
      "id": "moonshotai/kimi-k2.7-code",
      "orId": "moonshotai/kimi-k2.7-code",
      "vendor": "moonshotai",
      "model": "kimi-k2.7-code",
      "name": "MoonshotAI: Kimi K2.7 Code",
      "blurb": "MoonshotAI: Kimi K2.7 Code is a coding-focused model in Moonshot AI's Kimi K2 family, built to complete end-to-end programming tasks reliably over long contexts. It uses a native multimodal mixture-of-experts...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 1.2,
        "output": 7,
        "cacheRead": 0.32
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-06-12",
      "openWeights": {
        "huggingFace": "moonshotai/Kimi-K2.7-Code"
      },
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 151.15,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.879,
        "gpqaStderr": 0.0233,
        "aime": 0.956,
        "aimeStderr": 0.0311,
        "simpleQA": 0.365,
        "simpleQAStderr": 0.0152
      }
    },
    {
      "id": "moonshotai/kimi-k3",
      "orId": "moonshotai/kimi-k3",
      "vendor": "moonshotai",
      "model": "kimi-k3",
      "name": "MoonshotAI: Kimi K3",
      "blurb": "Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and is particularly strong at...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.384,
        "output": 32,
        "cacheRead": 0.364
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-07-16",
      "openWeights": {
        "huggingFace": "moonshotai/Kimi-K3"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 158.58,
        "accessibility": "Open weights (non-commercial)",
        "gpqa": 0.931,
        "gpqaStderr": 0.0149,
        "aime": 0.972,
        "aimeStderr": 0.0112,
        "simpleQA": 0.506,
        "simpleQAStderr": 0.0158
      }
    },
    {
      "id": "moonshotai/kimi-k3:batch",
      "orId": "moonshotai/kimi-k3:batch",
      "vendor": "moonshotai",
      "model": "kimi-k3:batch",
      "name": "MoonshotAI: Kimi K3 (batch)",
      "blurb": "Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and is particularly strong at...",
      "contextWindow": 1048576,
      "maxOutput": 16384,
      "cost": {
        "input": 4.56,
        "output": 22.8,
        "cacheRead": 0.456
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-07-16",
      "openWeights": {
        "huggingFace": "moonshotai/Kimi-K3"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "moonshotai/kimi-k3.1-pro",
      "orId": "moonshotai/kimi-k3.1-pro",
      "vendor": "moonshotai",
      "model": "kimi-k3.1-pro",
      "name": "MoonshotAI: Kimi K3.1 Pro",
      "blurb": "Kimi K3.1 Pro is Moonshot AI's flagship open-weight multimodal reasoning model and a post-trained successor to Kimi K3, with a 1M-token context window. It is built for long-horizon coding, 3D...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.4806,
        "output": 16.8046,
        "cacheRead": 0.4
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-10-23",
      "openWeights": {
        "huggingFace": "moonshotai/Kimi-K3.1-Pro"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 161.37,
        "accessibility": "Open weights (restricted use)"
      }
    },
    {
      "id": "moonshotai/kimi-k3.1-pro:batch",
      "orId": "moonshotai/kimi-k3.1-pro:batch",
      "vendor": "moonshotai",
      "model": "kimi-k3.1-pro:batch",
      "name": "MoonshotAI: Kimi K3.1 Pro (batch)",
      "blurb": "Kimi K3.1 Pro is Moonshot AI's flagship open-weight multimodal reasoning model and a post-trained successor to Kimi K3, with a 1M-token context window. It is built for long-horizon coding, 3D...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 1.4,
        "output": 12,
        "cacheRead": 0.2
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "max"
      },
      "tools": false,
      "released": "2026-10-23",
      "openWeights": {
        "huggingFace": "moonshotai/Kimi-K3.1-Pro"
      }
    },
    {
      "id": "morph/morph-v3-fast",
      "orId": "morph/morph-v3-fast",
      "vendor": "morph",
      "model": "morph-v3-fast",
      "name": "Morph: Morph V3 Fast",
      "blurb": "Morph's fastest apply model for code edits. ~10,500 tokens/sec with 96% accuracy for rapid code transformations. The model requires the prompt to be in the following format: <instruction>{instruction}</instruction> <code>{initial_code}</code> <update>{edit_snippet}</update>...",
      "contextWindow": 262144,
      "maxOutput": 131072,
      "cost": {
        "input": 1.6,
        "output": 2.4
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2025-07-07",
      "regions": [
        "us"
      ]
    },
    {
      "id": "morph/morph-v3-fast:batch",
      "orId": "morph/morph-v3-fast:batch",
      "vendor": "morph",
      "model": "morph-v3-fast:batch",
      "name": "Morph: Morph V3 Fast (batch)",
      "blurb": "Morph's fastest apply model for code edits. ~10,500 tokens/sec with 96% accuracy for rapid code transformations. The model requires the prompt to be in the following format: <instruction>{instruction}</instruction> <code>{initial_code}</code> <update>{edit_snippet}</update>...",
      "contextWindow": 81920,
      "maxOutput": 38000,
      "cost": {
        "input": 0.8,
        "output": 1.2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2025-07-07",
      "regions": [
        "us"
      ]
    },
    {
      "id": "morph/morph-v3-large",
      "orId": "morph/morph-v3-large",
      "vendor": "morph",
      "model": "morph-v3-large",
      "name": "Morph: Morph V3 Large",
      "blurb": "Morph's high-accuracy apply model for complex code edits. ~4,500 tokens/sec with 98% accuracy for precise code transformations. The model requires the prompt to be in the following format: <instruction>{instruction}</instruction> <code>{initial_code}</code>...",
      "contextWindow": 262144,
      "maxOutput": 131072,
      "cost": {
        "input": 1.8,
        "output": 3.8
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2025-07-07",
      "regions": [
        "us"
      ]
    },
    {
      "id": "morph/morph-v3-large:batch",
      "orId": "morph/morph-v3-large:batch",
      "vendor": "morph",
      "model": "morph-v3-large:batch",
      "name": "Morph: Morph V3 Large (batch)",
      "blurb": "Morph's high-accuracy apply model for complex code edits. ~4,500 tokens/sec with 98% accuracy for precise code transformations. The model requires the prompt to be in the following format: <instruction>{instruction}</instruction> <code>{initial_code}</code>...",
      "contextWindow": 262144,
      "maxOutput": 131072,
      "cost": {
        "input": 0.9,
        "output": 1.9
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2025-07-07",
      "regions": [
        "us"
      ]
    },
    {
      "id": "motif-technologies/motif-3",
      "orId": "motif-technologies/motif-3",
      "vendor": "motif-technologies",
      "model": "motif-3",
      "name": "Motif Technologies: Motif 3",
      "blurb": "Motif 3 is a Mixture-of-Experts language model from Motif Technologies, with 314B total and 13.2B active parameters. It is a general-purpose multilingual model for English and Korean, particularly strong at...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.6,
        "output": 2.4,
        "cacheRead": 0.12
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-11-02",
      "openWeights": {
        "huggingFace": "Motif-Technologies/Motif-3"
      }
    },
    {
      "id": "nex-agi/nex-n2.5-mini",
      "orId": "nex-agi/nex-n2.5-mini",
      "vendor": "nex-agi",
      "model": "nex-n2.5-mini",
      "name": "Nex AGI: Nex-N2.5-Mini",
      "blurb": "Nex-N2.5 is an agentic model built to turn goals into working, verified outcomes. Its core strength is agentic coding within a visual feedback loop: it can explore codebases, implement multi-file...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.05,
        "output": 0.2,
        "cacheRead": 0.005
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": false,
      "released": "2026-09-08",
      "openWeights": {
        "huggingFace": "nex-agi/Nex-N2.5-mini"
      }
    },
    {
      "id": "nex-agi/nex-n2.5-pro",
      "orId": "nex-agi/nex-n2.5-pro",
      "vendor": "nex-agi",
      "model": "nex-n2.5-pro",
      "name": "Nex AGI: Nex-N2.5-Pro",
      "blurb": "Nex-N2.5 is an agentic model built to turn goals into working, verified outcomes. Its core strength is agentic coding within a visual feedback loop: it can explore codebases, implement multi-file...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.15,
        "output": 0.5,
        "cacheRead": 0.03
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-08",
      "openWeights": {
        "huggingFace": "nex-agi/Nex-N2.5-Pro"
      }
    },
    {
      "id": "nex-agi/nex-n2.6",
      "orId": "nex-agi/nex-n2.6",
      "vendor": "nex-agi",
      "model": "nex-n2.6",
      "name": "Nex AGI: Nex-N2.6",
      "blurb": "Nex-N2.6 is an agentic model from Nex AGI, built on [GLM-5.3-Flash](https://huggingface.co/zai-org/GLM-5.3-Flash-BF16) with about 321B parameters. It is suited for computer use, coding, and long-horizon tasks that involve tools. Compared with...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.24,
        "output": 0.8,
        "cacheRead": 0.048
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-11-10",
      "openWeights": {
        "huggingFace": "nex-agi/Nex-N2.6"
      }
    },
    {
      "id": "nousresearch/hermes-4-405b",
      "orId": "nousresearch/hermes-4-405b",
      "vendor": "nousresearch",
      "model": "hermes-4-405b",
      "name": "Nous: Hermes 4 405B",
      "blurb": "Hermes 4 is a large-scale reasoning model built on Meta-Llama-3.1-405B and released by Nous Research. It introduces a hybrid reasoning mode, where the model can choose to deliberate internally with...",
      "contextWindow": 131072,
      "maxOutput": 117964,
      "cost": {
        "input": 2,
        "output": 6,
        "cacheRead": 0.4
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": false,
      "knowledgeCutoff": "2024-08-31",
      "released": "2025-08-26",
      "openWeights": {
        "huggingFace": "NousResearch/Hermes-4-405B"
      }
    },
    {
      "id": "nvidia/nemotron-3-nano-30b-a3b",
      "orId": "nvidia/nemotron-3-nano-30b-a3b",
      "vendor": "nvidia",
      "model": "nemotron-3-nano-30b-a3b",
      "name": "NVIDIA: Nemotron 3 Nano 30B A3B",
      "blurb": "NVIDIA Nemotron 3 Nano 30B A3B is a small language MoE model with highest compute efficiency and accuracy for developers to build specialized agentic AI systems. The model is fully...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.12,
        "output": 0.48,
        "cacheRead": 0.024
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2025-12-14",
      "openWeights": {
        "huggingFace": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16"
      }
    },
    {
      "id": "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free",
      "orId": "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free",
      "vendor": "nvidia",
      "model": "nemotron-3-nano-omni-30b-a3b-reasoning:free",
      "name": "NVIDIA: Nemotron 3 Nano Omni (free)",
      "blurb": "NVIDIA Nemotron™ 3 Nano Omni is a 30B-A3B open multimodal model designed to function as a perception and context sub-agent in enterprise agent systems. It accepts text, image, video, and...",
      "contextWindow": 256000,
      "maxOutput": 65536,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text",
        "audio",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-28",
      "openWeights": {
        "huggingFace": "nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16"
      }
    },
    {
      "id": "nvidia/nemotron-3-super-120b-a12b",
      "orId": "nvidia/nemotron-3-super-120b-a12b",
      "vendor": "nvidia",
      "model": "nemotron-3-super-120b-a12b",
      "name": "NVIDIA: Nemotron 3 Super",
      "blurb": "NVIDIA Nemotron 3 Super is a 120B-parameter open hybrid MoE model, activating just 12B parameters for maximum compute efficiency and accuracy in complex multi-agent applications. Built on a hybrid Mamba-Transformer...",
      "contextWindow": 262144,
      "maxOutput": 16384,
      "cost": {
        "input": 0.15,
        "output": 0.8
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-03-11",
      "openWeights": {
        "huggingFace": "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "nvidia/nemotron-3-ultra-550b-a55b",
      "orId": "nvidia/nemotron-3-ultra-550b-a55b",
      "vendor": "nvidia",
      "model": "nemotron-3-ultra-550b-a55b",
      "name": "NVIDIA: Nemotron 3 Ultra",
      "blurb": "NVIDIA Nemotron 3 Ultra is an open frontier-reasoning and orchestration model from NVIDIA, with 55B active parameters out of 550B total (MoE). Built on a hybrid Transformer-Mamba mixture-of-experts architecture, it...",
      "contextWindow": 262144,
      "maxOutput": 182520,
      "cost": {
        "input": 1.2,
        "output": 4.8,
        "cacheRead": 0.24
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-06-04",
      "openWeights": {
        "huggingFace": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 146.95,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.854,
        "gpqaStderr": 0.0252,
        "aime": 0.867,
        "aimeStderr": 0.0512
      }
    },
    {
      "id": "nvidia/nemotron-3-ultra-550b-a55b:free",
      "orId": "nvidia/nemotron-3-ultra-550b-a55b:free",
      "vendor": "nvidia",
      "model": "nemotron-3-ultra-550b-a55b:free",
      "name": "NVIDIA: Nemotron 3 Ultra (free)",
      "blurb": "NVIDIA Nemotron 3 Ultra is an open frontier-reasoning and orchestration model from NVIDIA, with 55B active parameters out of 550B total (MoE). Built on a hybrid Transformer-Mamba mixture-of-experts architecture, it...",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-06-04",
      "openWeights": {
        "huggingFace": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16"
      }
    },
    {
      "id": "nvidia/nemotron-3.5-content-safety",
      "orId": "nvidia/nemotron-3.5-content-safety",
      "vendor": "nvidia",
      "model": "nemotron-3.5-content-safety",
      "name": "NVIDIA: Nemotron 3.5 Content Safety",
      "blurb": "NVIDIA Nemotron 3.5 Content Safety is a compact 4B-parameter multimodal guardrail model from NVIDIA, fine-tuned from Google Gemma-3-4B. It moderates both inputs to and responses from LLMs and VLMs, accepting...",
      "contextWindow": 131072,
      "maxOutput": 117964,
      "cost": {
        "input": 0.4,
        "output": 0.4
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": false,
      "released": "2026-06-04",
      "openWeights": {
        "huggingFace": "nvidia/Nemotron-3.5-Content-Safety"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "nvidia/nemotron-3.5-lightning",
      "orId": "nvidia/nemotron-3.5-lightning",
      "vendor": "nvidia",
      "model": "nemotron-3.5-lightning",
      "name": "NVIDIA: Nemotron 3.5 Lightning",
      "blurb": "NVIDIA Nemotron 3.5 Lightning is an open mixture-of-experts model from NVIDIA, with 3B active parameters out of 30B total. It is suited for high-throughput agentic workloads and specialized tasks that...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.14,
        "output": 0.4,
        "cacheRead": 0.08
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-08-11",
      "openWeights": {
        "huggingFace": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "nvidia/nemotron-3.5-lightning:free",
      "orId": "nvidia/nemotron-3.5-lightning:free",
      "vendor": "nvidia",
      "model": "nemotron-3.5-lightning:free",
      "name": "NVIDIA: Nemotron 3.5 Lightning (free)",
      "blurb": "NVIDIA Nemotron 3.5 Lightning is an open mixture-of-experts model from NVIDIA, with 3B active parameters out of 30B total. It is suited for high-throughput agentic workloads and specialized tasks that...",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-08-11",
      "openWeights": {
        "huggingFace": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16"
      }
    },
    {
      "id": "nvidia/nemotron-3.5-super-vl",
      "orId": "nvidia/nemotron-3.5-super-vl",
      "vendor": "nvidia",
      "model": "nemotron-3.5-super-vl",
      "name": "NVIDIA: Nemotron 3.5 Super VL",
      "blurb": "NVIDIA Nemotron 3.5 Super VL is an open multimodal mixture-of-experts model from NVIDIA, with 12B active parameters out of 120B total. It takes text, images, documents, GUI screenshots, and video...",
      "contextWindow": 1048576,
      "maxOutput": 235929,
      "cost": {
        "input": 0.3,
        "output": 1,
        "cacheRead": 0.04
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-10-15",
      "openWeights": {
        "huggingFace": "nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "nvidia/nemotron-3.5-super-vl:free",
      "orId": "nvidia/nemotron-3.5-super-vl:free",
      "vendor": "nvidia",
      "model": "nemotron-3.5-super-vl:free",
      "name": "NVIDIA: Nemotron 3.5 Super VL (free)",
      "blurb": "NVIDIA Nemotron 3.5 Super VL is an open multimodal mixture-of-experts model from NVIDIA, with 12B active parameters out of 120B total. It takes text, images, documents, GUI screenshots, and video...",
      "contextWindow": 1000000,
      "maxOutput": 131072,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-10-15",
      "openWeights": {
        "huggingFace": "nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16"
      }
    },
    {
      "id": "nvidia/switchyard",
      "orId": "nvidia/switchyard",
      "vendor": "nvidia",
      "model": "switchyard",
      "name": "NVIDIA: Switchyard",
      "contextWindow": 1000000,
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2026-09-21"
    },
    {
      "id": "openai/gpt-3.5-turbo-0613",
      "orId": "openai/gpt-3.5-turbo-0613",
      "vendor": "openai",
      "model": "gpt-3.5-turbo-0613",
      "name": "OpenAI: GPT-3.5 Turbo (older v0613)",
      "blurb": "GPT-3.5 Turbo is OpenAI's fastest model. It can understand and generate natural language or code, and is optimized for chat and traditional completion tasks.\n\nTraining data up to Sep 2021.",
      "contextWindow": 4095,
      "maxOutput": 3685,
      "cost": {
        "input": 2,
        "output": 4
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2021-09-30",
      "released": "2024-01-25"
    },
    {
      "id": "openai/gpt-3.5-turbo-16k",
      "orId": "openai/gpt-3.5-turbo-16k",
      "vendor": "openai",
      "model": "gpt-3.5-turbo-16k",
      "name": "OpenAI: GPT-3.5 Turbo 16k",
      "blurb": "This model offers four times the context length of gpt-3.5-turbo, allowing it to support approximately 20 pages of text in a single request at a higher cost. Training data: up...",
      "contextWindow": 16385,
      "maxOutput": 4096,
      "cost": {
        "input": 6,
        "output": 8
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2021-09-30",
      "released": "2023-08-28"
    },
    {
      "id": "openai/gpt-4",
      "orId": "openai/gpt-4",
      "vendor": "openai",
      "model": "gpt-4",
      "name": "OpenAI: GPT-4",
      "blurb": "OpenAI's flagship model, GPT-4 is a large-scale multimodal language model capable of solving difficult problems with greater accuracy than previous models due to its broader general knowledge and advanced reasoning...",
      "contextWindow": 8191,
      "maxOutput": 4096,
      "cost": {
        "input": 60,
        "output": 120
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2021-09-30",
      "released": "2023-05-28"
    },
    {
      "id": "openai/gpt-4-turbo:batch",
      "orId": "openai/gpt-4-turbo:batch",
      "vendor": "openai",
      "model": "gpt-4-turbo:batch",
      "name": "OpenAI: GPT-4 Turbo (batch)",
      "blurb": "The latest GPT-4 Turbo model with vision capabilities. Vision requests can now use JSON mode and function calling.\n\nTraining data: up to December 2023.",
      "contextWindow": 128000,
      "maxOutput": 4096,
      "cost": {
        "input": 10,
        "output": 30,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-12-31",
      "released": "2024-04-09"
    },
    {
      "id": "openai/gpt-4.1",
      "orId": "openai/gpt-4.1",
      "vendor": "openai",
      "model": "gpt-4.1",
      "name": "OpenAI: GPT-4.1",
      "blurb": "GPT-4.1 is a flagship large language model optimized for advanced instruction following, real-world software engineering, and long-context reasoning. It supports a 1 million token context window and outperforms GPT-4o and...",
      "contextWindow": 1047576,
      "maxOutput": 32768,
      "cost": {
        "input": 4,
        "output": 16,
        "cacheRead": 1,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-04-14",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 136.15,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-4.1-mini",
      "orId": "openai/gpt-4.1-mini",
      "vendor": "openai",
      "model": "gpt-4.1-mini",
      "name": "OpenAI: GPT-4.1 Mini",
      "blurb": "GPT-4.1 Mini is a mid-sized model delivering performance competitive with GPT-4o at substantially lower latency and cost. It retains a 1 million token context window and scores 45.1% on hard...",
      "contextWindow": 1047576,
      "maxOutput": 32768,
      "cost": {
        "input": 0.8,
        "output": 3.2,
        "cacheRead": 0.2,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-04-14",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 134.2,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-4.1-mini:batch",
      "orId": "openai/gpt-4.1-mini:batch",
      "vendor": "openai",
      "model": "gpt-4.1-mini:batch",
      "name": "OpenAI: GPT-4.1 Mini (batch)",
      "blurb": "GPT-4.1 Mini is a mid-sized model delivering performance competitive with GPT-4o at substantially lower latency and cost. It retains a 1 million token context window and scores 45.1% on hard...",
      "contextWindow": 1047576,
      "maxOutput": 32768,
      "cost": {
        "input": 0.4,
        "output": 1.6,
        "cacheRead": 0.1,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-04-14"
    },
    {
      "id": "openai/gpt-4.1-nano",
      "orId": "openai/gpt-4.1-nano",
      "vendor": "openai",
      "model": "gpt-4.1-nano",
      "name": "OpenAI: GPT-4.1 Nano",
      "blurb": "For tasks that demand low latency, GPT‑4.1 nano is the fastest and cheapest model in the GPT-4.1 series. It delivers exceptional performance at a small size with its 1 million...",
      "contextWindow": 1047576,
      "maxOutput": 942818,
      "cost": {
        "input": 0.2,
        "output": 0.8,
        "cacheRead": 0.06,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-04-14",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 126.77,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-4.1-nano:batch",
      "orId": "openai/gpt-4.1-nano:batch",
      "vendor": "openai",
      "model": "gpt-4.1-nano:batch",
      "name": "OpenAI: GPT-4.1 Nano (batch)",
      "blurb": "For tasks that demand low latency, GPT‑4.1 nano is the fastest and cheapest model in the GPT-4.1 series. It delivers exceptional performance at a small size with its 1 million...",
      "contextWindow": 1047576,
      "maxOutput": 32768,
      "cost": {
        "input": 0.1,
        "output": 0.4,
        "cacheRead": 0.025,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-04-14"
    },
    {
      "id": "openai/gpt-4.1:batch",
      "orId": "openai/gpt-4.1:batch",
      "vendor": "openai",
      "model": "gpt-4.1:batch",
      "name": "OpenAI: GPT-4.1 (batch)",
      "blurb": "GPT-4.1 is a flagship large language model optimized for advanced instruction following, real-world software engineering, and long-context reasoning. It supports a 1 million token context window and outperforms GPT-4o and...",
      "contextWindow": 1047576,
      "maxOutput": 32768,
      "cost": {
        "input": 2,
        "output": 8,
        "cacheRead": 0.5,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-04-14"
    },
    {
      "id": "openai/gpt-4o",
      "orId": "openai/gpt-4o",
      "vendor": "openai",
      "model": "gpt-4o",
      "name": "OpenAI: GPT-4o",
      "blurb": "GPT-4o (\"o\" for \"omni\") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of [GPT-4 Turbo](/models/openai/gpt-4-turbo) while being twice as...",
      "contextWindow": 128000,
      "maxOutput": 16384,
      "cost": {
        "input": 5,
        "output": 20,
        "cacheRead": 2.5
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2024-05-13"
    },
    {
      "id": "openai/gpt-4o-2024-05-13",
      "orId": "openai/gpt-4o-2024-05-13",
      "vendor": "openai",
      "model": "gpt-4o-2024-05-13",
      "name": "OpenAI: GPT-4o (2024-05-13)",
      "blurb": "GPT-4o (\"o\" for \"omni\") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of [GPT-4 Turbo](/models/openai/gpt-4-turbo) while being twice as...",
      "contextWindow": 128000,
      "maxOutput": 4096,
      "cost": {
        "input": 10,
        "output": 30
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2024-05-13",
      "bench": {
        "source": "epoch",
        "gpqa": 0.489,
        "gpqaStderr": 0.0263,
        "mathLevel5": 0.51,
        "mathLevel5Stderr": 0.0112,
        "aime": 0.063,
        "aimeStderr": 0.0232
      }
    },
    {
      "id": "openai/gpt-4o-2024-08-06",
      "orId": "openai/gpt-4o-2024-08-06",
      "vendor": "openai",
      "model": "gpt-4o-2024-08-06",
      "name": "OpenAI: GPT-4o (2024-08-06)",
      "blurb": "The 2024-08-06 version of GPT-4o offers improved performance in structured outputs, with the ability to supply a JSON schema in the respone_format. Read more [here](https://openai.com/index/introducing-structured-outputs-in-the-api/). GPT-4o (\"o\" for \"omni\") is...",
      "contextWindow": 128000,
      "maxOutput": 16384,
      "cost": {
        "input": 5,
        "output": 20,
        "cacheRead": 2.5
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2024-08-06",
      "bench": {
        "source": "epoch",
        "gpqa": 0.492,
        "gpqaStderr": 0.026,
        "frontierMath": 0.003,
        "frontierMathStderr": 0.0034,
        "mathLevel5": 0.533,
        "mathLevel5Stderr": 0.0112,
        "aime": 0.064,
        "aimeStderr": 0.0256,
        "simpleQA": 0.26,
        "simpleQAStderr": 0.0139
      }
    },
    {
      "id": "openai/gpt-4o-2024-11-20",
      "orId": "openai/gpt-4o-2024-11-20",
      "vendor": "openai",
      "model": "gpt-4o-2024-11-20",
      "name": "OpenAI: GPT-4o (2024-11-20)",
      "blurb": "The 2024-11-20 version of GPT-4o offers a leveled-up creative writing ability with more natural, engaging, and tailored writing to improve relevance & readability. It’s also better at working with uploaded...",
      "contextWindow": 128000,
      "maxOutput": 16384,
      "cost": {
        "input": 5,
        "output": 20,
        "cacheRead": 2.5
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2024-11-20",
      "bench": {
        "source": "epoch",
        "gpqa": 0.479,
        "gpqaStderr": 0.0263,
        "sweBenchVerified": 0.31,
        "sweBenchVerifiedStderr": 0.021,
        "frontierMath": 0.003,
        "frontierMathStderr": 0.0034,
        "mathLevel5": 0.498,
        "mathLevel5Stderr": 0.0111,
        "aime": 0.063,
        "aimeStderr": 0.0242
      }
    },
    {
      "id": "openai/gpt-4o-mini",
      "orId": "openai/gpt-4o-mini",
      "vendor": "openai",
      "model": "gpt-4o-mini",
      "name": "OpenAI: GPT-4o-mini",
      "blurb": "GPT-4o mini is OpenAI's newest model after [GPT-4 Omni](/models/openai/gpt-4o), supporting both text and image inputs with text outputs. As their most advanced small model, it is many multiples more affordable...",
      "contextWindow": 128000,
      "maxOutput": 16384,
      "cost": {
        "input": 0.3,
        "output": 1.2,
        "cacheRead": 0.15
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2024-07-18",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 123.86,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-4o-mini-2024-07-18",
      "orId": "openai/gpt-4o-mini-2024-07-18",
      "vendor": "openai",
      "model": "gpt-4o-mini-2024-07-18",
      "name": "OpenAI: GPT-4o-mini (2024-07-18)",
      "blurb": "GPT-4o mini is OpenAI's newest model after [GPT-4 Omni](/models/openai/gpt-4o), supporting both text and image inputs with text outputs. As their most advanced small model, it is many multiples more affordable...",
      "contextWindow": 128000,
      "maxOutput": 16384,
      "cost": {
        "input": 0.3,
        "output": 1.2,
        "cacheRead": 0.15
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2024-07-18",
      "bench": {
        "source": "epoch",
        "gpqa": 0.377,
        "gpqaStderr": 0.0241,
        "mathLevel5": 0.526,
        "mathLevel5Stderr": 0.0113,
        "aime": 0.069,
        "aimeStderr": 0.0274,
        "simpleQA": 0.083,
        "simpleQAStderr": 0.0087
      }
    },
    {
      "id": "openai/gpt-4o-mini:batch",
      "orId": "openai/gpt-4o-mini:batch",
      "vendor": "openai",
      "model": "gpt-4o-mini:batch",
      "name": "OpenAI: GPT-4o-mini (batch)",
      "blurb": "GPT-4o mini is OpenAI's newest model after [GPT-4 Omni](/models/openai/gpt-4o), supporting both text and image inputs with text outputs. As their most advanced small model, it is many multiples more affordable...",
      "contextWindow": 128000,
      "maxOutput": 16384,
      "cost": {
        "input": 0.15,
        "output": 0.6,
        "cacheRead": 0.075,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2024-07-18"
    },
    {
      "id": "openai/gpt-4o:batch",
      "orId": "openai/gpt-4o:batch",
      "vendor": "openai",
      "model": "gpt-4o:batch",
      "name": "OpenAI: GPT-4o (batch)",
      "blurb": "GPT-4o (\"o\" for \"omni\") is OpenAI's latest AI model, supporting both text and image inputs with text outputs. It maintains the intelligence level of [GPT-4 Turbo](/models/openai/gpt-4-turbo) while being twice as...",
      "contextWindow": 128000,
      "maxOutput": 16384,
      "cost": {
        "input": 2.5,
        "output": 10,
        "cacheRead": 1.25,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2024-05-13"
    },
    {
      "id": "openai/gpt-5",
      "orId": "openai/gpt-5",
      "vendor": "openai",
      "model": "gpt-5",
      "name": "OpenAI: GPT-5",
      "blurb": "GPT-5 is OpenAI’s most advanced model, offering major improvements in reasoning, code quality, and user experience. It is optimized for complex tasks that require step-by-step reasoning, instruction following, and accuracy...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 2.5,
        "output": 20,
        "cacheRead": 0.25,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2024-09-30",
      "released": "2025-08-07",
      "retiresAt": "2026-12-11",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 150,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5-image",
      "orId": "openai/gpt-5-image",
      "vendor": "openai",
      "model": "gpt-5-image",
      "name": "OpenAI: GPT-5 Image",
      "blurb": "GPT-5 Image combines OpenAI's GPT-5 model with state-of-the-art image generation capabilities. It offers major improvements in reasoning, code quality, and user experience while incorporating GPT Image 1's superior instruction following,...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 20,
        "output": 20,
        "cacheRead": 2.5,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "image",
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": false,
      "released": "2025-10-14",
      "retiresAt": "2026-12-11"
    },
    {
      "id": "openai/gpt-5-image-mini",
      "orId": "openai/gpt-5-image-mini",
      "vendor": "openai",
      "model": "gpt-5-image-mini",
      "name": "OpenAI: GPT-5 Image Mini",
      "blurb": "GPT-5 Image Mini combines OpenAI's advanced language capabilities, powered by GPT-5 Mini, with GPT Image 1 Mini for efficient image generation. This natively multimodal model features superior instruction following, text...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 5,
        "output": 4,
        "cacheRead": 0.5,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "image",
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": false,
      "released": "2025-10-16",
      "retiresAt": "2026-12-01"
    },
    {
      "id": "openai/gpt-5-mini",
      "orId": "openai/gpt-5-mini",
      "vendor": "openai",
      "model": "gpt-5-mini",
      "name": "OpenAI: GPT-5 Mini",
      "blurb": "GPT-5 Mini is a compact version of GPT-5, designed to handle lighter-weight reasoning tasks. It provides the same instruction-following and safety-tuning benefits as GPT-5, but with reduced latency and cost....",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.5,
        "output": 4,
        "cacheRead": 0.05,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2024-05-31",
      "released": "2025-08-07",
      "retiresAt": "2026-12-11",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 146.09,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5-mini:batch",
      "orId": "openai/gpt-5-mini:batch",
      "vendor": "openai",
      "model": "gpt-5-mini:batch",
      "name": "OpenAI: GPT-5 Mini (batch)",
      "blurb": "GPT-5 Mini is a compact version of GPT-5, designed to handle lighter-weight reasoning tasks. It provides the same instruction-following and safety-tuning benefits as GPT-5, but with reduced latency and cost....",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.25,
        "output": 2,
        "cacheRead": 0.025,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2024-05-31",
      "released": "2025-08-07",
      "retiresAt": "2026-12-11"
    },
    {
      "id": "openai/gpt-5-nano",
      "orId": "openai/gpt-5-nano",
      "vendor": "openai",
      "model": "gpt-5-nano",
      "name": "OpenAI: GPT-5 Nano",
      "blurb": "GPT-5-Nano is the smallest and fastest variant in the GPT-5 system, optimized for developer tools, rapid interactions, and ultra-low latency environments. While limited in reasoning depth compared to its larger...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.1,
        "output": 0.8,
        "cacheRead": 0.01,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2024-05-31",
      "released": "2025-08-07",
      "retiresAt": "2026-12-11",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 139.1,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5-nano:batch",
      "orId": "openai/gpt-5-nano:batch",
      "vendor": "openai",
      "model": "gpt-5-nano:batch",
      "name": "OpenAI: GPT-5 Nano (batch)",
      "blurb": "GPT-5-Nano is the smallest and fastest variant in the GPT-5 system, optimized for developer tools, rapid interactions, and ultra-low latency environments. While limited in reasoning depth compared to its larger...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.05,
        "output": 0.4,
        "cacheRead": 0.005,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2024-05-31",
      "released": "2025-08-07",
      "retiresAt": "2026-12-11"
    },
    {
      "id": "openai/gpt-5-pro",
      "orId": "openai/gpt-5-pro",
      "vendor": "openai",
      "model": "gpt-5-pro",
      "name": "OpenAI: GPT-5 Pro",
      "blurb": "GPT-5 Pro is OpenAI’s most advanced model, offering major improvements in reasoning, code quality, and user experience. It is optimized for complex tasks that require step-by-step reasoning, instruction following, and...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 30,
        "output": 240,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "knowledgeCutoff": "2024-09-30",
      "released": "2025-10-06",
      "retiresAt": "2026-12-11",
      "bench": {
        "source": "epoch",
        "eci": 151.23,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5-pro:batch",
      "orId": "openai/gpt-5-pro:batch",
      "vendor": "openai",
      "model": "gpt-5-pro:batch",
      "name": "OpenAI: GPT-5 Pro (batch)",
      "blurb": "GPT-5 Pro is OpenAI’s most advanced model, offering major improvements in reasoning, code quality, and user experience. It is optimized for complex tasks that require step-by-step reasoning, instruction following, and...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 15,
        "output": 120,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "knowledgeCutoff": "2024-09-30",
      "released": "2025-10-06",
      "retiresAt": "2026-12-11"
    },
    {
      "id": "openai/gpt-5:batch",
      "orId": "openai/gpt-5:batch",
      "vendor": "openai",
      "model": "gpt-5:batch",
      "name": "OpenAI: GPT-5 (batch)",
      "blurb": "GPT-5 is OpenAI’s most advanced model, offering major improvements in reasoning, code quality, and user experience. It is optimized for complex tasks that require step-by-step reasoning, instruction following, and accuracy...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 1.25,
        "output": 10,
        "cacheRead": 0.125,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2024-09-30",
      "released": "2025-08-07",
      "retiresAt": "2026-12-11"
    },
    {
      "id": "openai/gpt-5.1",
      "orId": "openai/gpt-5.1",
      "vendor": "openai",
      "model": "gpt-5.1",
      "name": "OpenAI: GPT-5.1",
      "blurb": "GPT-5.1 is the latest frontier-grade model in the GPT-5 series, offering stronger general-purpose reasoning, improved instruction adherence, and a more natural conversational style compared to GPT-5. It uses adaptive reasoning...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 2.5,
        "output": 20,
        "cacheRead": 0.25,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-11-13",
      "regions": [
        "eu"
      ],
      "bench": {
        "source": "epoch",
        "eci": 150.94,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5.1-codex",
      "orId": "openai/gpt-5.1-codex",
      "vendor": "openai",
      "model": "gpt-5.1-codex",
      "name": "OpenAI: GPT-5.1-Codex",
      "blurb": "GPT-5.1-Codex is a specialized version of GPT-5.1 optimized for software engineering and coding workflows. It is designed for both interactive development sessions and long, independent execution of complex engineering tasks....",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 2.5,
        "output": 20,
        "cacheRead": 0.26,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-11-13"
    },
    {
      "id": "openai/gpt-5.1-codex-max",
      "orId": "openai/gpt-5.1-codex-max",
      "vendor": "openai",
      "model": "gpt-5.1-codex-max",
      "name": "OpenAI: GPT-5.1-Codex-Max",
      "blurb": "GPT-5.1-Codex-Max is OpenAI’s latest agentic coding model, designed for long-running, high-context software development tasks. It is based on an updated version of the 5.1 reasoning stack and trained on agentic...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 2.5,
        "output": 20,
        "cacheRead": 0.25,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-12-04"
    },
    {
      "id": "openai/gpt-5.1-codex-mini",
      "orId": "openai/gpt-5.1-codex-mini",
      "vendor": "openai",
      "model": "gpt-5.1-codex-mini",
      "name": "OpenAI: GPT-5.1-Codex-Mini",
      "blurb": "GPT-5.1-Codex-Mini is a smaller and faster version of GPT-5.1-Codex",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.5,
        "output": 4,
        "cacheRead": 0.06,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-11-13"
    },
    {
      "id": "openai/gpt-5.1:batch",
      "orId": "openai/gpt-5.1:batch",
      "vendor": "openai",
      "model": "gpt-5.1:batch",
      "name": "OpenAI: GPT-5.1 (batch)",
      "blurb": "GPT-5.1 is the latest frontier-grade model in the GPT-5 series, offering stronger general-purpose reasoning, improved instruction adherence, and a more natural conversational style compared to GPT-5. It uses adaptive reasoning...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 1.25,
        "output": 10,
        "cacheRead": 0.125,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-11-13"
    },
    {
      "id": "openai/gpt-5.2",
      "orId": "openai/gpt-5.2",
      "vendor": "openai",
      "model": "gpt-5.2",
      "name": "OpenAI: GPT-5.2",
      "blurb": "GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1. It uses adaptive reasoning to allocate computation dynamically, responding quickly...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 3.5,
        "output": 28,
        "cacheRead": 0.35,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-12-10",
      "bench": {
        "source": "epoch",
        "eci": 154.46,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5.2-chat",
      "orId": "openai/gpt-5.2-chat",
      "vendor": "openai",
      "model": "gpt-5.2-chat",
      "name": "OpenAI: GPT-5.2 Chat",
      "blurb": "GPT-5.2 Chat (AKA Instant) is the fast, lightweight member of the 5.2 family, optimized for low-latency chat while retaining strong general intelligence. It uses adaptive reasoning to selectively “think” on...",
      "contextWindow": 128000,
      "maxOutput": 32000,
      "cost": {
        "input": 3.5,
        "output": 28,
        "cacheRead": 0.35,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2025-12-10"
    },
    {
      "id": "openai/gpt-5.2-codex",
      "orId": "openai/gpt-5.2-codex",
      "vendor": "openai",
      "model": "gpt-5.2-codex",
      "name": "OpenAI: GPT-5.2-Codex",
      "blurb": "GPT-5.2-Codex is an upgraded version of GPT-5.1-Codex optimized for software engineering and coding workflows. It is designed for both interactive development sessions and long, independent execution of complex engineering tasks....",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 3.5,
        "output": 28,
        "cacheRead": 0.35,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-01-14"
    },
    {
      "id": "openai/gpt-5.2-pro",
      "orId": "openai/gpt-5.2-pro",
      "vendor": "openai",
      "model": "gpt-5.2-pro",
      "name": "OpenAI: GPT-5.2 Pro",
      "blurb": "GPT-5.2 Pro is OpenAI’s most advanced model, offering major improvements in agentic coding and long context performance over GPT-5 Pro. It is optimized for complex tasks that require step-by-step reasoning,...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 42,
        "output": 336,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-12-10",
      "bench": {
        "source": "epoch",
        "eci": 157.27,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5.2-pro:batch",
      "orId": "openai/gpt-5.2-pro:batch",
      "vendor": "openai",
      "model": "gpt-5.2-pro:batch",
      "name": "OpenAI: GPT-5.2 Pro (batch)",
      "blurb": "GPT-5.2 Pro is OpenAI’s most advanced model, offering major improvements in agentic coding and long context performance over GPT-5 Pro. It is optimized for complex tasks that require step-by-step reasoning,...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 21,
        "output": 168,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-12-10"
    },
    {
      "id": "openai/gpt-5.2:batch",
      "orId": "openai/gpt-5.2:batch",
      "vendor": "openai",
      "model": "gpt-5.2:batch",
      "name": "OpenAI: GPT-5.2 (batch)",
      "blurb": "GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1. It uses adaptive reasoning to allocate computation dynamically, responding quickly...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 1.75,
        "output": 14,
        "cacheRead": 0.175,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2025-12-10"
    },
    {
      "id": "openai/gpt-5.3-codex",
      "orId": "openai/gpt-5.3-codex",
      "vendor": "openai",
      "model": "gpt-5.3-codex",
      "name": "OpenAI: GPT-5.3-Codex",
      "blurb": "GPT-5.3-Codex is OpenAI’s most advanced agentic coding model, combining the frontier software engineering performance of GPT-5.2-Codex with the broader reasoning and professional knowledge capabilities of GPT-5.2. It achieves state-of-the-art results...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 3.5,
        "output": 28,
        "cacheRead": 0.35,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-02-24",
      "bench": {
        "source": "epoch",
        "sweBenchVerified": 0.748,
        "sweBenchVerifiedStderr": 0.0198
      }
    },
    {
      "id": "openai/gpt-5.4",
      "orId": "openai/gpt-5.4",
      "vendor": "openai",
      "model": "gpt-5.4",
      "name": "OpenAI: GPT-5.4",
      "blurb": "GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window (922K input, 128K output) with support for...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 5,
        "output": 30,
        "cacheRead": 0.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 10,
            "output": 45,
            "cacheRead": 1
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-03-05",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 157.85,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5.4-image-2",
      "orId": "openai/gpt-5.4-image-2",
      "vendor": "openai",
      "model": "gpt-5.4-image-2",
      "name": "OpenAI: GPT-5.4 Image 2",
      "blurb": "GPT-5.4 Image 2 combines OpenAI's GPT-5.4 model with state-of-the-art image generation capabilities from GPT Image 2. It enables rich multimodal workflows, allowing users to seamlessly move between reasoning, coding, and...",
      "contextWindow": 272000,
      "maxOutput": 128000,
      "cost": {
        "input": 16,
        "output": 30,
        "cacheRead": 4,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "image",
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": false,
      "released": "2026-04-21"
    },
    {
      "id": "openai/gpt-5.4-mini",
      "orId": "openai/gpt-5.4-mini",
      "vendor": "openai",
      "model": "gpt-5.4-mini",
      "name": "OpenAI: GPT-5.4 Mini",
      "blurb": "GPT-5.4 mini brings the core capabilities of GPT-5.4 to a faster, more efficient model optimized for high-throughput workloads. It supports text and image inputs with strong performance across reasoning, coding,...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 1.5,
        "output": 9,
        "cacheRead": 0.15,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-08-31",
      "released": "2026-03-17",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 149.34,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5.4-mini:batch",
      "orId": "openai/gpt-5.4-mini:batch",
      "vendor": "openai",
      "model": "gpt-5.4-mini:batch",
      "name": "OpenAI: GPT-5.4 Mini (batch)",
      "blurb": "GPT-5.4 mini brings the core capabilities of GPT-5.4 to a faster, more efficient model optimized for high-throughput workloads. It supports text and image inputs with strong performance across reasoning, coding,...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.75,
        "output": 4.5,
        "cacheRead": 0.075,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-08-31",
      "released": "2026-03-17"
    },
    {
      "id": "openai/gpt-5.4-nano",
      "orId": "openai/gpt-5.4-nano",
      "vendor": "openai",
      "model": "gpt-5.4-nano",
      "name": "OpenAI: GPT-5.4 Nano",
      "blurb": "GPT-5.4 nano is the most lightweight and cost-efficient variant of the GPT-5.4 family, optimized for speed-critical and high-volume tasks. It supports text and image inputs and is designed for low-latency...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.4,
        "output": 2.5,
        "cacheRead": 0.04,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-08-31",
      "released": "2026-03-17",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 146.4,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5.4-nano:batch",
      "orId": "openai/gpt-5.4-nano:batch",
      "vendor": "openai",
      "model": "gpt-5.4-nano:batch",
      "name": "OpenAI: GPT-5.4 Nano (batch)",
      "blurb": "GPT-5.4 nano is the most lightweight and cost-efficient variant of the GPT-5.4 family, optimized for speed-critical and high-volume tasks. It supports text and image inputs and is designed for low-latency...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.2,
        "output": 1.25,
        "cacheRead": 0.02,
        "webSearch": 0.02
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-08-31",
      "released": "2026-03-17"
    },
    {
      "id": "openai/gpt-5.4-pro",
      "orId": "openai/gpt-5.4-pro",
      "vendor": "openai",
      "model": "gpt-5.4-pro",
      "name": "OpenAI: GPT-5.4 Pro",
      "blurb": "GPT-5.4 Pro is OpenAI's most advanced model, building on GPT-5.4's unified architecture with enhanced reasoning capabilities for complex, high-stakes tasks. It features a 1M+ token context window (922K input, 128K...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 60,
        "output": 360,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 120,
            "output": 540
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-03-05",
      "bench": {
        "source": "epoch",
        "eci": 160.24,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5.4-pro:batch",
      "orId": "openai/gpt-5.4-pro:batch",
      "vendor": "openai",
      "model": "gpt-5.4-pro:batch",
      "name": "OpenAI: GPT-5.4 Pro (batch)",
      "blurb": "GPT-5.4 Pro is OpenAI's most advanced model, building on GPT-5.4's unified architecture with enhanced reasoning capabilities for complex, high-stakes tasks. It features a 1M+ token context window (922K input, 128K...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 30,
        "output": 180,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 60,
            "output": 270
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-03-05"
    },
    {
      "id": "openai/gpt-5.4:batch",
      "orId": "openai/gpt-5.4:batch",
      "vendor": "openai",
      "model": "gpt-5.4:batch",
      "name": "OpenAI: GPT-5.4 (batch)",
      "blurb": "GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window (922K input, 128K output) with support for...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 2.5,
        "output": 15,
        "cacheRead": 0.25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 5,
            "output": 22.5,
            "cacheRead": 0.5
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-03-05"
    },
    {
      "id": "openai/gpt-5.5",
      "orId": "openai/gpt-5.5",
      "vendor": "openai",
      "model": "gpt-5.5",
      "name": "OpenAI: GPT-5.5",
      "blurb": "GPT-5.5 is OpenAI’s frontier model designed for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. It features a 1M+ token...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 60,
        "cacheRead": 1,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 20,
            "output": 90,
            "cacheRead": 2
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-12-01",
      "released": "2026-04-24",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 160.1,
        "accessibility": "API access",
        "gpqa": 0.907,
        "gpqaStderr": 0.0175,
        "aime": 0.844,
        "aimeStderr": 0.0546,
        "simpleQA": 0.63,
        "simpleQAStderr": 0.0153
      }
    },
    {
      "id": "openai/gpt-5.5-pro",
      "orId": "openai/gpt-5.5-pro",
      "vendor": "openai",
      "model": "gpt-5.5-pro",
      "name": "OpenAI: GPT-5.5 Pro",
      "blurb": "GPT-5.5 Pro is OpenAI’s high-capability model optimized for deep reasoning and accuracy on complex, high-stakes workloads. It features a 1M+ token context window (922K input, 128K output) with support for...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 60,
        "output": 360,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 120,
            "output": 540
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-12-01",
      "released": "2026-04-24",
      "bench": {
        "source": "epoch",
        "eci": 162.7,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/gpt-5.5-pro:batch",
      "orId": "openai/gpt-5.5-pro:batch",
      "vendor": "openai",
      "model": "gpt-5.5-pro:batch",
      "name": "OpenAI: GPT-5.5 Pro (batch)",
      "blurb": "GPT-5.5 Pro is OpenAI’s high-capability model optimized for deep reasoning and accuracy on complex, high-stakes workloads. It features a 1M+ token context window (922K input, 128K output) with support for...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 30,
        "output": 180,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 60,
            "output": 270
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-12-01",
      "released": "2026-04-24"
    },
    {
      "id": "openai/gpt-5.5:batch",
      "orId": "openai/gpt-5.5:batch",
      "vendor": "openai",
      "model": "gpt-5.5:batch",
      "name": "OpenAI: GPT-5.5 (batch)",
      "blurb": "GPT-5.5 is OpenAI’s frontier model designed for complex professional workloads, building on GPT-5.4 with stronger reasoning, higher reliability, and improved token efficiency on hard tasks. It features a 1M+ token...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 5,
        "output": 30,
        "cacheRead": 0.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 10,
            "output": 45,
            "cacheRead": 1
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2025-12-01",
      "released": "2026-04-24"
    },
    {
      "id": "openai/gpt-5.6-luna",
      "orId": "openai/gpt-5.6-luna",
      "vendor": "openai",
      "model": "gpt-5.6-luna",
      "name": "OpenAI: GPT-5.6 Luna",
      "blurb": "GPT-5.6 Luna is a fast, cost-efficient model in OpenAI's GPT-5.6 series. It is suited for high-volume, latency-sensitive tasks such as chat, classification, and lightweight agentic workflows, providing capable reasoning for...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.4,
        "output": 2.4,
        "cacheRead": 0.04,
        "cacheWrite": 0.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 0.8,
            "output": 3.6,
            "cacheRead": 0.08
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 156.67,
        "accessibility": "API access",
        "gpqa": 0.916,
        "gpqaStderr": 0.0173,
        "aime": 0.983,
        "aimeStderr": 0.0117,
        "simpleQA": 0.41,
        "simpleQAStderr": 0.0156
      }
    },
    {
      "id": "openai/gpt-5.6-luna-pro",
      "orId": "openai/gpt-5.6-luna-pro",
      "vendor": "openai",
      "model": "gpt-5.6-luna-pro",
      "name": "OpenAI: GPT-5.6 Luna Pro",
      "blurb": "GPT-5.6 Luna Pro is the same underlying model as GPT-5.6 Luna, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.4,
        "output": 2.4,
        "cacheRead": 0.04,
        "cacheWrite": 0.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 0.8,
            "output": 3.6,
            "cacheRead": 0.08
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09",
      "regions": [
        "eu"
      ]
    },
    {
      "id": "openai/gpt-5.6-luna-pro:batch",
      "orId": "openai/gpt-5.6-luna-pro:batch",
      "vendor": "openai",
      "model": "gpt-5.6-luna-pro:batch",
      "name": "OpenAI: GPT-5.6 Luna Pro (batch)",
      "blurb": "GPT-5.6 Luna Pro is the same underlying model as GPT-5.6 Luna, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.2,
        "output": 1.2,
        "cacheRead": 0.02,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 0.4,
            "output": 1.8,
            "cacheRead": 0.04
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09"
    },
    {
      "id": "openai/gpt-5.6-luna:batch",
      "orId": "openai/gpt-5.6-luna:batch",
      "vendor": "openai",
      "model": "gpt-5.6-luna:batch",
      "name": "OpenAI: GPT-5.6 Luna (batch)",
      "blurb": "GPT-5.6 Luna is a fast, cost-efficient model in OpenAI's GPT-5.6 series. It is suited for high-volume, latency-sensitive tasks such as chat, classification, and lightweight agentic workflows, providing capable reasoning for...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.2,
        "output": 1.2,
        "cacheRead": 0.02,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 0.4,
            "output": 1.8,
            "cacheRead": 0.04
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09"
    },
    {
      "id": "openai/gpt-5.6-sol",
      "orId": "openai/gpt-5.6-sol",
      "vendor": "openai",
      "model": "gpt-5.6-sol",
      "name": "OpenAI: GPT-5.6 Sol",
      "blurb": "GPT-5.6 Sol is the flagship model in OpenAI's GPT-5.6 series. It is suited for complex reasoning, coding, and agentic workflows, and is particularly strong at command-line and multi-step coding tasks...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 8,
        "output": 40,
        "cacheRead": 0.8,
        "cacheWrite": 10,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 16,
            "output": 60,
            "cacheRead": 1.6
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 162.99,
        "accessibility": "API access",
        "gpqa": 0.935,
        "gpqaStderr": 0.0157,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.697,
        "simpleQAStderr": 0.0145
      }
    },
    {
      "id": "openai/gpt-5.6-sol-pro",
      "orId": "openai/gpt-5.6-sol-pro",
      "vendor": "openai",
      "model": "gpt-5.6-sol-pro",
      "name": "OpenAI: GPT-5.6 Sol Pro",
      "blurb": "GPT-5.6 Sol Pro is the same underlying model as GPT-5.6 Sol, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 8,
        "output": 40,
        "cacheRead": 0.8,
        "cacheWrite": 10,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 16,
            "output": 60,
            "cacheRead": 1.6
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09",
      "regions": [
        "eu"
      ]
    },
    {
      "id": "openai/gpt-5.6-sol-pro:batch",
      "orId": "openai/gpt-5.6-sol-pro:batch",
      "vendor": "openai",
      "model": "gpt-5.6-sol-pro:batch",
      "name": "OpenAI: GPT-5.6 Sol Pro (batch)",
      "blurb": "GPT-5.6 Sol Pro is the same underlying model as GPT-5.6 Sol, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.4,
        "cacheWrite": 5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 8,
            "output": 30,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09"
    },
    {
      "id": "openai/gpt-5.6-sol:batch",
      "orId": "openai/gpt-5.6-sol:batch",
      "vendor": "openai",
      "model": "gpt-5.6-sol:batch",
      "name": "OpenAI: GPT-5.6 Sol (batch)",
      "blurb": "GPT-5.6 Sol is the flagship model in OpenAI's GPT-5.6 series. It is suited for complex reasoning, coding, and agentic workflows, and is particularly strong at command-line and multi-step coding tasks...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.4,
        "cacheWrite": 5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 8,
            "output": 30,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09"
    },
    {
      "id": "openai/gpt-5.6-terra",
      "orId": "openai/gpt-5.6-terra",
      "vendor": "openai",
      "model": "gpt-5.6-terra",
      "name": "OpenAI: GPT-5.6 Terra",
      "blurb": "GPT-5.6 Terra is a balanced model in OpenAI's GPT-5.6 series, positioned between the flagship Sol tier and the cost-efficient Luna tier. It is suited for everyday coding, reasoning, and agentic...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 24,
        "cacheRead": 0.4,
        "cacheWrite": 5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 8,
            "output": 36,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 160.73,
        "accessibility": "API access",
        "gpqa": 0.933,
        "gpqaStderr": 0.0154,
        "aime": 0.997,
        "aimeStderr": 0.0028,
        "simpleQA": 0.432,
        "simpleQAStderr": 0.0157
      }
    },
    {
      "id": "openai/gpt-5.6-terra-pro",
      "orId": "openai/gpt-5.6-terra-pro",
      "vendor": "openai",
      "model": "gpt-5.6-terra-pro",
      "name": "OpenAI: GPT-5.6 Terra Pro",
      "blurb": "GPT-5.6 Terra Pro is the same underlying model as GPT-5.6 Terra, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 24,
        "cacheRead": 0.4,
        "cacheWrite": 5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 8,
            "output": 36,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09",
      "regions": [
        "eu"
      ]
    },
    {
      "id": "openai/gpt-5.6-terra-pro:batch",
      "orId": "openai/gpt-5.6-terra-pro:batch",
      "vendor": "openai",
      "model": "gpt-5.6-terra-pro:batch",
      "name": "OpenAI: GPT-5.6 Terra Pro (batch)",
      "blurb": "GPT-5.6 Terra Pro is the same underlying model as GPT-5.6 Terra, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 12,
        "cacheRead": 0.2,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 4,
            "output": 18,
            "cacheRead": 0.4
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09"
    },
    {
      "id": "openai/gpt-5.6-terra:batch",
      "orId": "openai/gpt-5.6-terra:batch",
      "vendor": "openai",
      "model": "gpt-5.6-terra:batch",
      "name": "OpenAI: GPT-5.6 Terra (batch)",
      "blurb": "GPT-5.6 Terra is a balanced model in OpenAI's GPT-5.6 series, positioned between the flagship Sol tier and the cost-efficient Luna tier. It is suited for everyday coding, reasoning, and agentic...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 12,
        "cacheRead": 0.2,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 4,
            "output": 18,
            "cacheRead": 0.4
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2026-02-16",
      "released": "2026-07-09"
    },
    {
      "id": "openai/gpt-6-astra",
      "orId": "openai/gpt-6-astra",
      "vendor": "openai",
      "model": "gpt-6-astra",
      "name": "OpenAI: GPT-6 Astra",
      "blurb": "GPT-6 Astra is OpenAI's flagship model for demanding end-to-end work. It is suited for advanced analysis, software engineering, deep research, scientific work, and document creation, with particular strengths in long-horizon...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 20,
        "output": 100,
        "cacheRead": 2,
        "cacheWrite": 25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 40,
            "output": 150,
            "cacheRead": 4
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-04",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 169,
        "accessibility": "API access",
        "gpqa": 0.958,
        "gpqaStderr": 0.0137,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.756,
        "simpleQAStderr": 0.0136
      }
    },
    {
      "id": "openai/gpt-6-astra-pro",
      "orId": "openai/gpt-6-astra-pro",
      "vendor": "openai",
      "model": "gpt-6-astra-pro",
      "name": "OpenAI: GPT-6 Astra Pro",
      "blurb": "GPT-6 Astra Pro is the same underlying model as GPT-6 Astra, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 20,
        "output": 100,
        "cacheRead": 2,
        "cacheWrite": 25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 40,
            "output": 150,
            "cacheRead": 4
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-04",
      "regions": [
        "us"
      ]
    },
    {
      "id": "openai/gpt-6-astra-pro:batch",
      "orId": "openai/gpt-6-astra-pro:batch",
      "vendor": "openai",
      "model": "gpt-6-astra-pro:batch",
      "name": "OpenAI: GPT-6 Astra Pro (batch)",
      "blurb": "GPT-6 Astra Pro is the same underlying model as GPT-6 Astra, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 50,
        "cacheRead": 1,
        "cacheWrite": 12.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 20,
            "output": 75,
            "cacheRead": 2
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-04"
    },
    {
      "id": "openai/gpt-6-astra:batch",
      "orId": "openai/gpt-6-astra:batch",
      "vendor": "openai",
      "model": "gpt-6-astra:batch",
      "name": "OpenAI: GPT-6 Astra (batch)",
      "blurb": "GPT-6 Astra is OpenAI's flagship model for demanding end-to-end work. It is suited for advanced analysis, software engineering, deep research, scientific work, and document creation, with particular strengths in long-horizon...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 50,
        "cacheRead": 1,
        "cacheWrite": 12.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 20,
            "output": 75,
            "cacheRead": 2
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-04"
    },
    {
      "id": "openai/gpt-6-luna",
      "orId": "openai/gpt-6-luna",
      "vendor": "openai",
      "model": "gpt-6-luna",
      "name": "OpenAI: GPT-6 Luna",
      "blurb": "GPT-6 Luna is the fast, cost-efficient model in OpenAI's GPT-6 series, positioned below GPT-6 Sol. It is suited for high-volume and latency-sensitive workloads such as chat, classification, and lightweight agentic...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.2,
        "output": 1,
        "cacheRead": 0.02,
        "cacheWrite": 0.25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 0.4,
            "output": 1.5,
            "cacheRead": 0.04
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-22",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 156.41,
        "accessibility": "API access",
        "gpqa": 0.905,
        "gpqaStderr": 0.0185,
        "aime": 0.989,
        "aimeStderr": 0.0067,
        "simpleQA": 0.414,
        "simpleQAStderr": 0.0156
      }
    },
    {
      "id": "openai/gpt-6-luna-pro",
      "orId": "openai/gpt-6-luna-pro",
      "vendor": "openai",
      "model": "gpt-6-luna-pro",
      "name": "OpenAI: GPT-6 Luna Pro",
      "blurb": "GPT-6 Luna Pro is the same underlying model as GPT-6 Luna, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.2,
        "output": 1,
        "cacheRead": 0.02,
        "cacheWrite": 0.25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 0.4,
            "output": 1.5,
            "cacheRead": 0.04
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-22",
      "regions": [
        "eu",
        "us"
      ]
    },
    {
      "id": "openai/gpt-6-luna-pro:batch",
      "orId": "openai/gpt-6-luna-pro:batch",
      "vendor": "openai",
      "model": "gpt-6-luna-pro:batch",
      "name": "OpenAI: GPT-6 Luna Pro (batch)",
      "blurb": "GPT-6 Luna Pro is the same underlying model as GPT-6 Luna, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.1,
        "output": 0.5,
        "cacheRead": 0.01,
        "cacheWrite": 0.125,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 0.2,
            "output": 0.75,
            "cacheRead": 0.02
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-22"
    },
    {
      "id": "openai/gpt-6-luna:batch",
      "orId": "openai/gpt-6-luna:batch",
      "vendor": "openai",
      "model": "gpt-6-luna:batch",
      "name": "OpenAI: GPT-6 Luna (batch)",
      "blurb": "GPT-6 Luna is the fast, cost-efficient model in OpenAI's GPT-6 series, positioned below GPT-6 Sol. It is suited for high-volume and latency-sensitive workloads such as chat, classification, and lightweight agentic...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 0.1,
        "output": 0.5,
        "cacheRead": 0.01,
        "cacheWrite": 0.125,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 0.2,
            "output": 0.75,
            "cacheRead": 0.02
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-22"
    },
    {
      "id": "openai/gpt-6-sol",
      "orId": "openai/gpt-6-sol",
      "vendor": "openai",
      "model": "gpt-6-sol",
      "name": "OpenAI: GPT-6 Sol",
      "blurb": "GPT-6 Sol is the cost-efficient high-end model in OpenAI's GPT-6 series, positioned below the flagship GPT-6 Astra and above the fast GPT-6 Luna tier. It is suited for demanding professional...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.4,
        "cacheWrite": 5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 8,
            "output": 30,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-22",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 163.75,
        "accessibility": "API access",
        "gpqa": 0.943,
        "gpqaStderr": 0.0149,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.607,
        "simpleQAStderr": 0.0155
      }
    },
    {
      "id": "openai/gpt-6-sol-pro",
      "orId": "openai/gpt-6-sol-pro",
      "vendor": "openai",
      "model": "gpt-6-sol-pro",
      "name": "OpenAI: GPT-6 Sol Pro",
      "blurb": "GPT-6 Sol Pro is the same underlying model as GPT-6 Sol, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.4,
        "cacheWrite": 5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 8,
            "output": 30,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-22",
      "regions": [
        "eu",
        "us"
      ]
    },
    {
      "id": "openai/gpt-6-sol-pro:batch",
      "orId": "openai/gpt-6-sol-pro:batch",
      "vendor": "openai",
      "model": "gpt-6-sol-pro:batch",
      "name": "OpenAI: GPT-6 Sol Pro (batch)",
      "blurb": "GPT-6 Sol Pro is the same underlying model as GPT-6 Sol, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.2,
        "cacheWrite": 2.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 4,
            "output": 15,
            "cacheRead": 0.4
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-22"
    },
    {
      "id": "openai/gpt-6-sol:batch",
      "orId": "openai/gpt-6-sol:batch",
      "vendor": "openai",
      "model": "gpt-6-sol:batch",
      "name": "OpenAI: GPT-6 Sol (batch)",
      "blurb": "GPT-6 Sol is the cost-efficient high-end model in OpenAI's GPT-6 series, positioned below the flagship GPT-6 Astra and above the fast GPT-6 Luna tier. It is suited for demanding professional...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.2,
        "cacheWrite": 2.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 4,
            "output": 15,
            "cacheRead": 0.4
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-22"
    },
    {
      "id": "openai/gpt-6.1-sol",
      "orId": "openai/gpt-6.1-sol",
      "vendor": "openai",
      "model": "gpt-6.1-sol",
      "name": "OpenAI: GPT-6.1 Sol",
      "blurb": "GPT-6.1 Sol is an upgrade to GPT-6 Sol from OpenAI, positioned below the flagship GPT-6 Astra in the GPT-6 series. It is suited for agentic coding, computer use, document-heavy professional...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.2,
        "cacheWrite": 5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 8,
            "output": 30,
            "cacheRead": 0.4
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-29",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 168.48,
        "accessibility": "API access",
        "gpqa": 0.954,
        "gpqaStderr": 0.0138,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.739,
        "simpleQAStderr": 0.0139
      }
    },
    {
      "id": "openai/gpt-6.1-sol-pro",
      "orId": "openai/gpt-6.1-sol-pro",
      "vendor": "openai",
      "model": "gpt-6.1-sol-pro",
      "name": "OpenAI: GPT-6.1 Sol Pro",
      "blurb": "GPT-6.1 Sol Pro is the same underlying model as GPT-6.1 Sol, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 20,
        "cacheRead": 0.2,
        "cacheWrite": 5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 8,
            "output": 30,
            "cacheRead": 0.4
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-29",
      "regions": [
        "eu",
        "us"
      ]
    },
    {
      "id": "openai/gpt-6.1-sol-pro:batch",
      "orId": "openai/gpt-6.1-sol-pro:batch",
      "vendor": "openai",
      "model": "gpt-6.1-sol-pro:batch",
      "name": "OpenAI: GPT-6.1 Sol Pro (batch)",
      "blurb": "GPT-6.1 Sol Pro is the same underlying model as GPT-6.1 Sol, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.1,
        "cacheWrite": 2.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 4,
            "output": 15,
            "cacheRead": 0.2
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-29"
    },
    {
      "id": "openai/gpt-6.1-sol:batch",
      "orId": "openai/gpt-6.1-sol:batch",
      "vendor": "openai",
      "model": "gpt-6.1-sol:batch",
      "name": "OpenAI: GPT-6.1 Sol (batch)",
      "blurb": "GPT-6.1 Sol is an upgrade to GPT-6 Sol from OpenAI, positioned below the flagship GPT-6 Astra in the GPT-6 series. It is suited for agentic coding, computer use, document-heavy professional...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.1,
        "cacheWrite": 2.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 4,
            "output": 15,
            "cacheRead": 0.2
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-29"
    },
    {
      "id": "openai/gpt-6.1-terra",
      "orId": "openai/gpt-6.1-terra",
      "vendor": "openai",
      "model": "gpt-6.1-terra",
      "name": "OpenAI: GPT-6.1 Terra",
      "blurb": "GPT-6.1 Terra is the mid-tier model in OpenAI's GPT-6.1 series, priced below GPT-6.1 Sol. It is suited for everyday agentic, coding, and writing work, is particularly strong at understanding screenshots...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.1,
        "cacheWrite": 2.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 4,
            "output": 15,
            "cacheRead": 0.2
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-26",
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 167.67,
        "accessibility": "API access",
        "gpqa": 0.943,
        "gpqaStderr": 0.0149,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.573,
        "simpleQAStderr": 0.0156
      }
    },
    {
      "id": "openai/gpt-6.1-terra-pro",
      "orId": "openai/gpt-6.1-terra-pro",
      "vendor": "openai",
      "model": "gpt-6.1-terra-pro",
      "name": "OpenAI: GPT-6.1 Terra Pro",
      "blurb": "GPT-6.1 Terra Pro is the same underlying model as GPT-6.1 Terra, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 10,
        "cacheRead": 0.1,
        "cacheWrite": 2.5,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 4,
            "output": 15,
            "cacheRead": 0.2
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-26",
      "regions": [
        "eu",
        "us"
      ]
    },
    {
      "id": "openai/gpt-6.1-terra-pro:batch",
      "orId": "openai/gpt-6.1-terra-pro:batch",
      "vendor": "openai",
      "model": "gpt-6.1-terra-pro:batch",
      "name": "OpenAI: GPT-6.1 Terra Pro (batch)",
      "blurb": "GPT-6.1 Terra Pro is the same underlying model as GPT-6.1 Terra, served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. **Cost note:** pro mode spends far more...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 1,
        "output": 5,
        "cacheRead": 0.05,
        "cacheWrite": 1.25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 2,
            "output": 7.5,
            "cacheRead": 0.1
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-26"
    },
    {
      "id": "openai/gpt-6.1-terra:batch",
      "orId": "openai/gpt-6.1-terra:batch",
      "vendor": "openai",
      "model": "gpt-6.1-terra:batch",
      "name": "OpenAI: GPT-6.1 Terra (batch)",
      "blurb": "GPT-6.1 Terra is the mid-tier model in OpenAI's GPT-6.1 series, priced below GPT-6.1 Sol. It is suited for everyday agentic, coding, and writing work, is particularly strong at understanding screenshots...",
      "contextWindow": 1050000,
      "maxOutput": 128000,
      "cost": {
        "input": 1,
        "output": 5,
        "cacheRead": 0.05,
        "cacheWrite": 1.25,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 2,
            "output": 7.5,
            "cacheRead": 0.1
          }
        ]
      },
      "input": [
        "file",
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-26"
    },
    {
      "id": "openai/gpt-audio",
      "orId": "openai/gpt-audio",
      "vendor": "openai",
      "model": "gpt-audio",
      "name": "OpenAI: GPT Audio",
      "blurb": "The gpt-audio model is OpenAI's first generally available audio model. The new snapshot features an upgraded decoder for more natural sounding voices and maintains better voice consistency. Audio is priced...",
      "contextWindow": 128000,
      "maxOutput": 16384,
      "cost": {
        "input": 5,
        "output": 20,
        "audioIn": 64,
        "audioOut": 128
      },
      "input": [
        "text",
        "audio"
      ],
      "output": [
        "text",
        "audio"
      ],
      "tools": true,
      "released": "2026-01-19"
    },
    {
      "id": "openai/gpt-audio-mini",
      "orId": "openai/gpt-audio-mini",
      "vendor": "openai",
      "model": "gpt-audio-mini",
      "name": "OpenAI: GPT Audio Mini",
      "blurb": "A cost-efficient version of GPT Audio. The new snapshot features an upgraded decoder for more natural sounding voices and maintains better voice consistency. Input is priced at $0.60 per million...",
      "contextWindow": 128000,
      "maxOutput": 16384,
      "cost": {
        "input": 1.2,
        "output": 4.8,
        "audioIn": 1.2,
        "audioOut": 4.8
      },
      "input": [
        "text",
        "audio"
      ],
      "output": [
        "text",
        "audio"
      ],
      "tools": true,
      "released": "2026-01-19"
    },
    {
      "id": "openai/gpt-chat-latest",
      "orId": "openai/gpt-chat-latest",
      "vendor": "openai",
      "model": "gpt-chat-latest",
      "name": "OpenAI: GPT Chat Latest",
      "blurb": "GPT Chat Latest points to OpenAI's stable API alias `chat-latest` that always resolves to the latest Instant chat model used in ChatGPT. As OpenAI rolls out new Instant model updates...",
      "contextWindow": 400000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 60,
        "cacheRead": 1,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2026-05-05"
    },
    {
      "id": "openai/gpt-oss-120b",
      "orId": "openai/gpt-oss-120b",
      "vendor": "openai",
      "model": "gpt-oss-120b",
      "name": "OpenAI: gpt-oss-120b",
      "blurb": "gpt-oss-120b is an open-weight, 117B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases. It activates 5.1B parameters per forward pass and is optimized...",
      "contextWindow": 131072,
      "maxOutput": 65536,
      "cost": {
        "input": 0.3,
        "output": 1.2,
        "cacheRead": 0.15
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-08-05",
      "openWeights": {
        "huggingFace": "openai/gpt-oss-120b"
      },
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 140.16,
        "accessibility": "Open weights (unrestricted)"
      }
    },
    {
      "id": "openai/gpt-oss-120b:batch",
      "orId": "openai/gpt-oss-120b:batch",
      "vendor": "openai",
      "model": "gpt-oss-120b:batch",
      "name": "OpenAI: gpt-oss-120b (batch)",
      "blurb": "gpt-oss-120b is an open-weight, 117B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases. It activates 5.1B parameters per forward pass and is optimized...",
      "contextWindow": 131072,
      "maxOutput": 117964,
      "cost": {
        "input": 0.0592,
        "output": 0.272
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-08-05",
      "openWeights": {
        "huggingFace": "openai/gpt-oss-120b"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "openai/gpt-oss-20b",
      "orId": "openai/gpt-oss-20b",
      "vendor": "openai",
      "model": "gpt-oss-20b",
      "name": "OpenAI: gpt-oss-20b",
      "blurb": "gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for...",
      "contextWindow": 131072,
      "maxOutput": 65536,
      "cost": {
        "input": 0.15,
        "output": 0.6,
        "cacheRead": 0.075
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-08-05",
      "openWeights": {
        "huggingFace": "openai/gpt-oss-20b"
      },
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 134.97,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.608,
        "gpqaStderr": 0.0249,
        "aime": 0.653,
        "aimeStderr": 0.0593
      }
    },
    {
      "id": "openai/gpt-oss-20b:batch",
      "orId": "openai/gpt-oss-20b:batch",
      "vendor": "openai",
      "model": "gpt-oss-20b:batch",
      "name": "OpenAI: gpt-oss-20b (batch)",
      "blurb": "gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for...",
      "contextWindow": 131072,
      "maxOutput": 117964,
      "cost": {
        "input": 0.048,
        "output": 0.224
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-08-05",
      "openWeights": {
        "huggingFace": "openai/gpt-oss-20b"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "openai/gpt-oss-safeguard-20b",
      "orId": "openai/gpt-oss-safeguard-20b",
      "vendor": "openai",
      "model": "gpt-oss-safeguard-20b",
      "name": "OpenAI: gpt-oss-safeguard-20b",
      "blurb": "gpt-oss-safeguard-20b is a safety reasoning model from OpenAI built upon gpt-oss-20b. This open-weight, 21B-parameter Mixture-of-Experts (MoE) model offers lower latency for safety tasks like content classification, LLM filtering, and trust...",
      "contextWindow": 131072,
      "maxOutput": 65536,
      "cost": {
        "input": 0.15,
        "output": 0.6,
        "cacheRead": 0.075
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2025-10-29",
      "openWeights": {
        "huggingFace": "openai/gpt-oss-safeguard-20b"
      }
    },
    {
      "id": "openai/o3",
      "orId": "openai/o3",
      "vendor": "openai",
      "model": "o3",
      "name": "OpenAI: o3",
      "blurb": "o3 is a well-rounded and powerful model across domains. It sets a new standard for math, science, coding, and visual reasoning tasks. It also excels at technical writing and instruction-following....",
      "contextWindow": 200000,
      "maxOutput": 100000,
      "cost": {
        "input": 4,
        "output": 16,
        "cacheRead": 1,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-04-16",
      "retiresAt": "2026-12-11",
      "bench": {
        "source": "epoch",
        "eci": 147.05,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/o3-mini:batch",
      "orId": "openai/o3-mini:batch",
      "vendor": "openai",
      "model": "o3-mini:batch",
      "name": "OpenAI: o3 Mini (batch)",
      "blurb": "OpenAI o3-mini is a cost-efficient language model optimized for STEM reasoning tasks, particularly excelling in science, mathematics, and coding. This model supports the `reasoning_effort` parameter, which can be set to...",
      "contextWindow": 200000,
      "maxOutput": 100000,
      "cost": {
        "input": 1.1,
        "output": 4.4,
        "cacheRead": 0.55,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "knowledgeCutoff": "2023-10-31",
      "released": "2025-01-31"
    },
    {
      "id": "openai/o3-pro",
      "orId": "openai/o3-pro",
      "vendor": "openai",
      "model": "o3-pro",
      "name": "OpenAI: o3 Pro",
      "blurb": "The o-series of models are trained with reinforcement learning to think before they answer and perform complex reasoning. The o3-pro model uses more compute to think harder and provide consistently...",
      "contextWindow": 200000,
      "maxOutput": 100000,
      "cost": {
        "input": 40,
        "output": 160,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "file",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-06-10",
      "retiresAt": "2026-12-11",
      "bench": {
        "source": "epoch",
        "eci": 147.46,
        "accessibility": "API access"
      }
    },
    {
      "id": "openai/o3:batch",
      "orId": "openai/o3:batch",
      "vendor": "openai",
      "model": "o3:batch",
      "name": "OpenAI: o3 (batch)",
      "blurb": "o3 is a well-rounded and powerful model across domains. It sets a new standard for math, science, coding, and visual reasoning tasks. It also excels at technical writing and instruction-following....",
      "contextWindow": 200000,
      "maxOutput": 100000,
      "cost": {
        "input": 2,
        "output": 8,
        "cacheRead": 0.5,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-04-16",
      "retiresAt": "2026-12-11"
    },
    {
      "id": "openai/o4-mini:batch",
      "orId": "openai/o4-mini:batch",
      "vendor": "openai",
      "model": "o4-mini:batch",
      "name": "OpenAI: o4 Mini (batch)",
      "blurb": "OpenAI o4-mini is a compact reasoning model in the o-series, optimized for fast, cost-efficient performance while retaining strong multimodal and agentic capabilities. It supports tool use and demonstrates competitive reasoning...",
      "contextWindow": 200000,
      "maxOutput": 100000,
      "cost": {
        "input": 1.1,
        "output": 4.4,
        "cacheRead": 0.275,
        "webSearch": 0.02
      },
      "input": [
        "image",
        "text",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-04-16"
    },
    {
      "id": "openrouter/auto",
      "orId": "openrouter/auto",
      "vendor": "openrouter",
      "model": "auto",
      "name": "Auto Router",
      "contextWindow": 2000000,
      "input": [
        "text",
        "image",
        "audio",
        "file",
        "video"
      ],
      "output": [
        "text",
        "image"
      ],
      "tools": true,
      "released": "2026-11-05"
    },
    {
      "id": "openrouter/auto-beta",
      "orId": "openrouter/auto-beta",
      "vendor": "openrouter",
      "model": "auto-beta",
      "name": "Auto Router (Beta)",
      "blurb": "The experimental version of our Auto Router where we test new improvements. Use it to get the latest and greatest version of our general purpose auto router, but expect beta...",
      "contextWindow": 2000000,
      "maxOutput": 64000,
      "input": [
        "text",
        "image",
        "audio",
        "file",
        "video"
      ],
      "output": [
        "text",
        "image"
      ],
      "tools": true,
      "released": "2026-07-17"
    },
    {
      "id": "openrouter/auto-code",
      "orId": "openrouter/auto-code",
      "vendor": "openrouter",
      "model": "auto-code",
      "name": "Auto Code",
      "contextWindow": 2000000,
      "maxOutput": 64000,
      "input": [
        "text",
        "image",
        "audio",
        "file",
        "video"
      ],
      "output": [
        "text",
        "image"
      ],
      "tools": true,
      "released": "2026-11-05"
    },
    {
      "id": "openrouter/bodybuilder",
      "orId": "openrouter/bodybuilder",
      "vendor": "openrouter",
      "model": "bodybuilder",
      "name": "Body Builder (beta)",
      "contextWindow": 128000,
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2025-12-05"
    },
    {
      "id": "openrouter/free",
      "orId": "openrouter/free",
      "vendor": "openrouter",
      "model": "free",
      "name": "Free Models Router",
      "contextWindow": 200000,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2026-02-01"
    },
    {
      "id": "openrouter/fusion",
      "orId": "openrouter/fusion",
      "vendor": "openrouter",
      "model": "fusion",
      "name": "Fusion",
      "blurb": "Fusion turns your prompt into a small multi-model deliberation. A panel of expert models (see below) analyzes your prompt in parallel with web search and web fetch enabled, then a...",
      "contextWindow": 1000000,
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2026-06-13"
    },
    {
      "id": "openrouter/pareto-code",
      "orId": "openrouter/pareto-code",
      "vendor": "openrouter",
      "model": "pareto-code",
      "name": "Pareto Code Router",
      "blurb": "The Pareto Router maintains a tiered shortlist of strong coding models, ranked by [Artificial Analysis](https://artificialanalysis.ai/) coding percentiles. Set min_coding_score between 0 and 1 on the pareto-router plugin to control how...",
      "contextWindow": 2000000,
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2026-04-21"
    },
    {
      "id": "particle-ai/particle-1",
      "orId": "particle-ai/particle-1",
      "vendor": "particle-ai",
      "model": "particle-1",
      "name": "Particle AI: Particle-1",
      "blurb": "Particle-1 is a token-efficient agentic coding model from Particle AI, post-trained on DeepSeek V4.1 Flash. It is suited for software engineering work such as terminal workflows, codebase changes, and data...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0.12,
        "output": 1,
        "cacheRead": 0.006
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-10-28",
      "regions": [
        "us"
      ]
    },
    {
      "id": "particle-ai/particle-1:free",
      "orId": "particle-ai/particle-1:free",
      "vendor": "particle-ai",
      "model": "particle-1:free",
      "name": "Particle AI: Particle-1 (free)",
      "blurb": "Particle-1 is a token-efficient agentic coding model from Particle AI, post-trained on DeepSeek V4.1 Flash. It is suited for software engineering work such as terminal workflows, codebase changes, and data...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-10-28",
      "regions": [
        "us"
      ]
    },
    {
      "id": "perceptron/perceptron-mk1",
      "orId": "perceptron/perceptron-mk1",
      "vendor": "perceptron",
      "model": "perceptron-mk1",
      "name": "Perceptron: Perceptron Mk1",
      "blurb": "Perceptron Mk1 (Mark One) is Perceptron's highest-quality vision-language model for video and embodied reasoning.** It accepts image and video inputs paired with natural language queries, and produces detailed visual understanding...",
      "contextWindow": 32768,
      "maxOutput": 8192,
      "cost": {
        "input": 0.3,
        "output": 3
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": false,
      "released": "2026-05-12"
    },
    {
      "id": "perceptron/perceptron-mk1.5",
      "orId": "perceptron/perceptron-mk1.5",
      "vendor": "perceptron",
      "model": "perceptron-mk1.5",
      "name": "Perceptron: Perceptron Mk1.5",
      "blurb": "Perceptron Mk1.5 is Perceptron's embodied reasoning model for physical agents. It accepts text, image, video, and audio input, and answers with text plus optional structured annotations: points, boxes, polygons, tracks,...",
      "contextWindow": 36864,
      "maxOutput": 8192,
      "cost": {
        "input": 0.3,
        "output": 3
      },
      "input": [
        "text",
        "image",
        "video",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-25"
    },
    {
      "id": "perplexity/sonar",
      "orId": "perplexity/sonar",
      "vendor": "perplexity",
      "model": "sonar",
      "name": "Perplexity: Sonar",
      "blurb": "Sonar is lightweight, affordable, fast, and simple to use — now featuring citations and the ability to customize sources. It is designed for companies seeking to integrate lightweight question-and-answer features...",
      "contextWindow": 127072,
      "maxOutput": 114364,
      "cost": {
        "input": 2,
        "output": 2,
        "webSearch": 0.01
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2025-01-27"
    },
    {
      "id": "perplexity/sonar-deep-research",
      "orId": "perplexity/sonar-deep-research",
      "vendor": "perplexity",
      "model": "sonar-deep-research",
      "name": "Perplexity: Sonar Deep Research",
      "blurb": "Sonar Deep Research is a research-focused model designed for multi-step retrieval, synthesis, and reasoning across complex topics. It autonomously searches, reads, and evaluates sources, refining its approach as it gathers...",
      "contextWindow": 128000,
      "maxOutput": 115200,
      "cost": {
        "input": 4,
        "output": 16,
        "webSearch": 0.01
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": false,
      "released": "2025-03-07"
    },
    {
      "id": "perplexity/sonar-pro",
      "orId": "perplexity/sonar-pro",
      "vendor": "perplexity",
      "model": "sonar-pro",
      "name": "Perplexity: Sonar Pro",
      "blurb": "Note: Sonar Pro pricing includes Perplexity search pricing. See [details here](https://docs.perplexity.ai/guides/pricing#detailed-pricing-breakdown-for-sonar-reasoning-pro-and-sonar-pro) For enterprises seeking more advanced capabilities, the Sonar Pro API can handle in-depth, mu",
      "contextWindow": 200000,
      "maxOutput": 8000,
      "cost": {
        "input": 6,
        "output": 30,
        "webSearch": 0.01
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2025-03-07"
    },
    {
      "id": "perplexity/sonar-pro-search",
      "orId": "perplexity/sonar-pro-search",
      "vendor": "perplexity",
      "model": "sonar-pro-search",
      "name": "Perplexity: Sonar Pro Search",
      "contextWindow": 200000,
      "maxOutput": 8000,
      "cost": {
        "input": 6,
        "output": 30,
        "webSearch": 0.036
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": false,
      "released": "2025-10-30"
    },
    {
      "id": "perplexity/sonar-reasoning-pro",
      "orId": "perplexity/sonar-reasoning-pro",
      "vendor": "perplexity",
      "model": "sonar-reasoning-pro",
      "name": "Perplexity: Sonar Reasoning Pro",
      "blurb": "Note: Sonar Pro pricing includes Perplexity search pricing. See [details here](https://docs.perplexity.ai/guides/pricing#detailed-pricing-breakdown-for-sonar-reasoning-pro-and-sonar-pro) Sonar Reasoning Pro is a premier reasoning model powered by DeepSeek R1 with Chain of Thought",
      "contextWindow": 128000,
      "maxOutput": 115200,
      "cost": {
        "input": 4,
        "output": 16,
        "webSearch": 0.01
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": false,
      "released": "2025-03-07"
    },
    {
      "id": "prism-ml/ternary-bonsai-2-27b",
      "orId": "prism-ml/ternary-bonsai-2-27b",
      "vendor": "prism-ml",
      "model": "ternary-bonsai-2-27b",
      "name": "PrismML: Ternary Bonsai 2 27B",
      "blurb": "Bonsai 2 27B is a 27B-parameter reasoning model from PrismML derived from Qwen3.8-27B. It supports coding, mathematics, tool calling, and image understanding with a 262K-token context window. Ternary compression shrinks...",
      "contextWindow": 262144,
      "maxOutput": 32768,
      "cost": {
        "input": 0.15,
        "output": 1,
        "cacheRead": 0.075
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "medium",
          "none"
        ],
        "defaultEffort": "xhigh"
      },
      "tools": true,
      "released": "2026-09-18",
      "openWeights": {
        "huggingFace": "prism-ml/Ternary-Bonsai-2-27B-gguf"
      }
    },
    {
      "id": "prism-ml/ternary-bonsai-2-9b",
      "orId": "prism-ml/ternary-bonsai-2-9b",
      "vendor": "prism-ml",
      "model": "ternary-bonsai-2-9b",
      "name": "PrismML: Ternary Bonsai 2 9B",
      "blurb": "Ternary Bonsai 2 9B is PrismML’s compact, open-weight language model for conversational AI, reasoning, and tool use. Built on Qwen3.5, it uses ternary weights packed into 2 bits for efficient...",
      "contextWindow": 262144,
      "maxOutput": 65536,
      "cost": {
        "input": 0.16,
        "output": 0.26,
        "cacheRead": 0.08
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-10-16",
      "openWeights": {
        "huggingFace": "prism-ml/Ternary-Bonsai-2-9B-gguf"
      }
    },
    {
      "id": "qwen/qwen-2.5-72b-instruct",
      "orId": "qwen/qwen-2.5-72b-instruct",
      "vendor": "qwen",
      "model": "qwen-2.5-72b-instruct",
      "name": "Qwen2.5 72B Instruct",
      "blurb": "Qwen2.5 72B is the latest series of Qwen large language models. Qwen2.5 brings the following improvements upon Qwen2: - Significantly more knowledge and has greatly improved capabilities in coding and...",
      "contextWindow": 32768,
      "maxOutput": 16384,
      "cost": {
        "input": 0.72,
        "output": 0.8
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2024-06-30",
      "released": "2024-09-19",
      "openWeights": {
        "huggingFace": "Qwen/Qwen2.5-72B-Instruct"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "gpqa": 0.491,
        "gpqaStderr": 0.0269,
        "mathLevel5": 0.632,
        "mathLevel5Stderr": 0.0107,
        "aime": 0.081,
        "aimeStderr": 0.0261
      }
    },
    {
      "id": "qwen/qwen-2.5-coder-32b-instruct",
      "orId": "qwen/qwen-2.5-coder-32b-instruct",
      "vendor": "qwen",
      "model": "qwen-2.5-coder-32b-instruct",
      "name": "Qwen2.5 Coder 32B Instruct",
      "blurb": "Qwen2.5-Coder is the latest series of Code-Specific Qwen large language models (formerly known as CodeQwen). Qwen2.5-Coder brings the following improvements upon CodeQwen1.5: - Significantly improvements in **code generation**, **code reasoning**...",
      "contextWindow": 32768,
      "maxOutput": 29491,
      "cost": {
        "input": 1.32,
        "output": 2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-06-30",
      "released": "2024-11-11",
      "openWeights": {
        "huggingFace": "Qwen/Qwen2.5-Coder-32B-Instruct"
      }
    },
    {
      "id": "qwen/qwen-plus",
      "orId": "qwen/qwen-plus",
      "vendor": "qwen",
      "model": "qwen-plus",
      "name": "Qwen: Qwen-Plus",
      "blurb": "Qwen-Plus, based on the Qwen2.5 foundation model, is a 131K context model with a balanced performance, speed, and cost combination.",
      "contextWindow": 1000000,
      "maxOutput": 32768,
      "cost": {
        "input": 0.52,
        "output": 1.56,
        "cacheRead": 0.104,
        "cacheWrite": 0.65,
        "tiers": [
          {
            "minPromptTokens": 256000,
            "input": 1.56,
            "output": 4.68,
            "cacheRead": 0.312
          }
        ]
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-02-01"
    },
    {
      "id": "qwen/qwen2.5-vl-72b-instruct",
      "orId": "qwen/qwen2.5-vl-72b-instruct",
      "vendor": "qwen",
      "model": "qwen2.5-vl-72b-instruct",
      "name": "Qwen: Qwen2.5 VL 72B Instruct",
      "blurb": "Qwen2.5-VL is proficient in recognizing common objects such as flowers, birds, fish, and insects. It is also highly capable of analyzing texts, charts, icons, graphics, and layouts within images.",
      "contextWindow": 128000,
      "maxOutput": 115200,
      "cost": {
        "input": 1.6,
        "output": 2,
        "cacheRead": 0.8
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-02-01",
      "openWeights": {
        "huggingFace": "Qwen/Qwen2.5-VL-72B-Instruct"
      }
    },
    {
      "id": "qwen/qwen3-14b",
      "orId": "qwen/qwen3-14b",
      "vendor": "qwen",
      "model": "qwen3-14b",
      "name": "Qwen: Qwen3 14B",
      "blurb": "Qwen3-14B is a dense 14.8B parameter causal language model from the Qwen3 series, designed for both complex reasoning and efficient dialogue. It supports seamless switching between a \"thinking\" mode for...",
      "contextWindow": 40960,
      "maxOutput": 16384,
      "cost": {
        "input": 0.24,
        "output": 0.48
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-04-28",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-14B"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 136.14,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.638,
        "gpqaStderr": 0.0293,
        "aime": 0.664,
        "aimeStderr": 0.0619
      }
    },
    {
      "id": "qwen/qwen3-235b-a22b-2507",
      "orId": "qwen/qwen3-235b-a22b-2507",
      "vendor": "qwen",
      "model": "qwen3-235b-a22b-2507",
      "name": "Qwen: Qwen3 235B A22B Instruct 2507",
      "blurb": "Qwen3-235B-A22B-Instruct-2507 is a multilingual, instruction-tuned mixture-of-experts language model based on the Qwen3-235B architecture, with 22B active parameters per forward pass. It is optimized for general-purpose text generation, including instruction following,...",
      "contextWindow": 262144,
      "maxOutput": 117964,
      "cost": {
        "input": 0.28,
        "output": 1.6,
        "cacheRead": 0.1
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-06-30",
      "released": "2025-07-21",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-235B-A22B-Instruct-2507"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "qwen/qwen3-235b-a22b-thinking-2507",
      "orId": "qwen/qwen3-235b-a22b-thinking-2507",
      "vendor": "qwen",
      "model": "qwen3-235b-a22b-thinking-2507",
      "name": "Qwen: Qwen3 235B A22B Thinking 2507",
      "blurb": "Qwen3-235B-A22B-Thinking-2507 is a high-performance, open-weight Mixture-of-Experts (MoE) language model optimized for complex reasoning tasks. It activates 22B of its 235B parameters per forward pass and natively supports up to 262,144...",
      "contextWindow": 128000,
      "maxOutput": 16384,
      "cost": {
        "input": 0.9,
        "output": 7
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "knowledgeCutoff": "2025-06-30",
      "released": "2025-07-25",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-235B-A22B-Thinking-2507"
      },
      "bench": {
        "source": "epoch",
        "gpqa": 0.801,
        "gpqaStderr": 0.026,
        "aime": 0.867,
        "aimeStderr": 0.0512,
        "simpleQA": 0.404,
        "simpleQAStderr": 0.0155
      }
    },
    {
      "id": "qwen/qwen3-30b-a3b",
      "orId": "qwen/qwen3-30b-a3b",
      "vendor": "qwen",
      "model": "qwen3-30b-a3b",
      "name": "Qwen: Qwen3 30B A3B",
      "blurb": "Qwen3, the latest generation in the Qwen large language model series, features both dense and mixture-of-experts (MoE) architectures to excel in reasoning, multilingual support, and advanced agent tasks. Its unique...",
      "contextWindow": 40960,
      "maxOutput": 16384,
      "cost": {
        "input": 0.24,
        "output": 1
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-04-28",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-30B-A3B"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 134.14,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.617,
        "gpqaStderr": 0.0297,
        "aime": 0.628,
        "aimeStderr": 0.0621
      }
    },
    {
      "id": "qwen/qwen3-30b-a3b-instruct-2507",
      "orId": "qwen/qwen3-30b-a3b-instruct-2507",
      "vendor": "qwen",
      "model": "qwen3-30b-a3b-instruct-2507",
      "name": "Qwen: Qwen3 30B A3B Instruct 2507",
      "blurb": "Qwen3-30B-A3B-Instruct-2507 is a 30.5B-parameter mixture-of-experts language model from Qwen, with 3.3B active parameters per inference. It operates in non-thinking mode and is designed for high-quality instruction following, multilingual understanding, and...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.2,
        "output": 0.6,
        "cacheRead": 0.2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-06-30",
      "released": "2025-07-29",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-30B-A3B-Instruct-2507"
      },
      "bench": {
        "source": "epoch",
        "gpqa": 0.556,
        "gpqaStderr": 0.0288,
        "aime": 0.622,
        "aimeStderr": 0.0651
      }
    },
    {
      "id": "qwen/qwen3-32b",
      "orId": "qwen/qwen3-32b",
      "vendor": "qwen",
      "model": "qwen3-32b",
      "name": "Qwen: Qwen3 32B",
      "blurb": "Qwen3-32B is a dense 32.8B parameter causal language model from the Qwen3 series, optimized for both complex reasoning and efficient dialogue. It supports seamless switching between a \"thinking\" mode for...",
      "contextWindow": 131072,
      "maxOutput": 16384,
      "cost": {
        "input": 0.16,
        "output": 0.56
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-04-28",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-32B"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 136.89,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.657,
        "gpqaStderr": 0.0279,
        "aime": 0.669,
        "aimeStderr": 0.0604
      }
    },
    {
      "id": "qwen/qwen3-coder",
      "orId": "qwen/qwen3-coder",
      "vendor": "qwen",
      "model": "qwen3-coder",
      "name": "Qwen: Qwen3 Coder 480B A35B",
      "blurb": "Qwen3-Coder-480B-A35B-Instruct is a Mixture-of-Experts (MoE) code generation model developed by the Qwen team. It is optimized for agentic coding tasks such as function calling, tool use, and long-context reasoning over...",
      "contextWindow": 262144,
      "maxOutput": 65536,
      "cost": {
        "input": 0.6,
        "output": 2,
        "cacheRead": 0.2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-06-30",
      "released": "2025-07-23",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-Coder-480B-A35B-Instruct"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "qwen/qwen3-coder-30b-a3b-instruct",
      "orId": "qwen/qwen3-coder-30b-a3b-instruct",
      "vendor": "qwen",
      "model": "qwen3-coder-30b-a3b-instruct",
      "name": "Qwen: Qwen3 Coder 30B A3B Instruct",
      "blurb": "Qwen3-Coder-30B-A3B-Instruct is a 30.5B parameter Mixture-of-Experts (MoE) model with 128 experts (8 active per forward pass), designed for advanced code generation, repository-scale understanding, and agentic tool use. Built on the...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.3,
        "output": 1.2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-06-30",
      "released": "2025-07-31",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-Coder-30B-A3B-Instruct"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "qwen/qwen3-coder-flash",
      "orId": "qwen/qwen3-coder-flash",
      "vendor": "qwen",
      "model": "qwen3-coder-flash",
      "name": "Qwen: Qwen3 Coder Flash",
      "blurb": "Qwen3 Coder Flash is Alibaba's fast and cost efficient version of their proprietary Qwen3 Coder Plus. It is a powerful coding agent model specializing in autonomous programming via tool calling...",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "cost": {
        "input": 0.39,
        "output": 1.95,
        "cacheRead": 0.078,
        "cacheWrite": 0.4875,
        "tiers": [
          {
            "minPromptTokens": 32000,
            "input": 0.65,
            "output": 3.25,
            "cacheRead": 0.13
          },
          {
            "minPromptTokens": 128000,
            "input": 1.04,
            "output": 5.2,
            "cacheRead": 0.208
          }
        ]
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-06-30",
      "released": "2025-09-17"
    },
    {
      "id": "qwen/qwen3-coder-next",
      "orId": "qwen/qwen3-coder-next",
      "vendor": "qwen",
      "model": "qwen3-coder-next",
      "name": "Qwen: Qwen3 Coder Next",
      "blurb": "Qwen3-Coder-Next is an open-weight causal language model optimized for coding agents and local development workflows. It uses a sparse MoE design with 80B total parameters and only 3B activated per...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.24,
        "output": 1.6,
        "cacheRead": 0.14
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2026-02-04",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-Coder-Next"
      }
    },
    {
      "id": "qwen/qwen3-next-80b-a3b-instruct",
      "orId": "qwen/qwen3-next-80b-a3b-instruct",
      "vendor": "qwen",
      "model": "qwen3-next-80b-a3b-instruct",
      "name": "Qwen: Qwen3 Next 80B A3B Instruct",
      "blurb": "Qwen3-Next-80B-A3B-Instruct is an instruction-tuned chat model in the Qwen3-Next series optimized for fast, stable responses without “thinking” traces. It targets complex tasks across reasoning, code generation, knowledge QA, and multilingual...",
      "contextWindow": 262144,
      "maxOutput": 16384,
      "cost": {
        "input": 0.18,
        "output": 2.2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-09-30",
      "released": "2025-09-11",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-Next-80B-A3B-Instruct"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "qwen/qwen3-vl-235b-a22b-instruct",
      "orId": "qwen/qwen3-vl-235b-a22b-instruct",
      "vendor": "qwen",
      "model": "qwen3-vl-235b-a22b-instruct",
      "name": "Qwen: Qwen3 VL 235B A22B Instruct",
      "blurb": "Qwen3-VL-235B-A22B Instruct is an open-weight multimodal model that unifies strong text generation with visual understanding across images and video. The Instruct model targets general vision-language use (VQA, document parsing, chart/table...",
      "contextWindow": 262144,
      "maxOutput": 32768,
      "cost": {
        "input": 0.42,
        "output": 3.8,
        "cacheRead": 0.2
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-09-23",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-VL-235B-A22B-Instruct"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "qwen/qwen3-vl-30b-a3b-instruct",
      "orId": "qwen/qwen3-vl-30b-a3b-instruct",
      "vendor": "qwen",
      "model": "qwen3-vl-30b-a3b-instruct",
      "name": "Qwen: Qwen3 VL 30B A3B Instruct",
      "blurb": "Qwen3-VL-30B-A3B-Instruct is a multimodal model that unifies strong text generation with visual understanding for images and videos. Its Instruct variant optimizes instruction-following for general multimodal tasks. It excels in perception...",
      "contextWindow": 262144,
      "maxOutput": 16384,
      "cost": {
        "input": 0.3,
        "output": 1.2
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-10-06",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-VL-30B-A3B-Instruct"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "qwen/qwen3-vl-8b-instruct",
      "orId": "qwen/qwen3-vl-8b-instruct",
      "vendor": "qwen",
      "model": "qwen3-vl-8b-instruct",
      "name": "Qwen: Qwen3 VL 8B Instruct",
      "blurb": "Qwen3-VL-8B-Instruct is a multimodal vision-language model from the Qwen3-VL series, built for high-fidelity understanding and reasoning across text, images, and video. It features improved multimodal fusion with Interleaved-MRoPE for long-horizon...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.5,
        "output": 1.5,
        "cacheRead": 0.24
      },
      "input": [
        "image",
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2025-10-14",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3-VL-8B-Instruct"
      }
    },
    {
      "id": "qwen/qwen3.5-122b-a10b",
      "orId": "qwen/qwen3.5-122b-a10b",
      "vendor": "qwen",
      "model": "qwen3.5-122b-a10b",
      "name": "Qwen: Qwen3.5-122B-A10B",
      "blurb": "The Qwen3.5 122B-A10B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. In terms of...",
      "contextWindow": 262144,
      "maxOutput": 65536,
      "cost": {
        "input": 0.52,
        "output": 4.16
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-02-25",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.5-122B-A10B"
      }
    },
    {
      "id": "qwen/qwen3.5-27b",
      "orId": "qwen/qwen3.5-27b",
      "vendor": "qwen",
      "model": "qwen3.5-27b",
      "name": "Qwen: Qwen3.5-27B",
      "blurb": "The Qwen3.5 27B native vision-language Dense model incorporates a linear attention mechanism, delivering fast response times while balancing inference speed and performance. Its overall capabilities are comparable to those of...",
      "contextWindow": 262144,
      "maxOutput": 81920,
      "cost": {
        "input": 0.52,
        "output": 5.2
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-02-25",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.5-27B"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "qwen/qwen3.5-35b-a3b",
      "orId": "qwen/qwen3.5-35b-a3b",
      "vendor": "qwen",
      "model": "qwen3.5-35b-a3b",
      "name": "Qwen: Qwen3.5-35B-A3B",
      "blurb": "The Qwen3.5 Series 35B-A3B is a native vision-language model designed with a hybrid architecture that integrates linear attention mechanisms and a sparse mixture-of-experts model, achieving higher inference efficiency. Its overall...",
      "contextWindow": 262144,
      "maxOutput": 81920,
      "cost": {
        "input": 0.28,
        "output": 2,
        "cacheRead": 0.1
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-02-25",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.5-35B-A3B"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 143.98,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.835,
        "gpqaStderr": 0.0218,
        "aime": 0.7,
        "aimeStderr": 0.06
      }
    },
    {
      "id": "qwen/qwen3.5-397b-a17b",
      "orId": "qwen/qwen3.5-397b-a17b",
      "vendor": "qwen",
      "model": "qwen3.5-397b-a17b",
      "name": "Qwen: Qwen3.5 397B A17B",
      "blurb": "The Qwen3.5 series 397B-A17B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. It delivers...",
      "contextWindow": 262144,
      "maxOutput": 81920,
      "cost": {
        "input": 0.9,
        "output": 6,
        "cacheRead": 0.44
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-02-16",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.5-397B-A17B"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 147.9,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.864,
        "gpqaStderr": 0.0245,
        "aime": 0.889,
        "aimeStderr": 0.0474
      }
    },
    {
      "id": "qwen/qwen3.5-9b",
      "orId": "qwen/qwen3.5-9b",
      "vendor": "qwen",
      "model": "qwen3.5-9b",
      "name": "Qwen: Qwen3.5-9B",
      "blurb": "Qwen3.5-9B is a multimodal foundation model from the Qwen3.5 family, designed to deliver strong reasoning, coding, and visual understanding in an efficient 9B-parameter architecture. It uses a unified vision-language design...",
      "contextWindow": 262144,
      "maxOutput": 40960,
      "cost": {
        "input": 0.138,
        "output": 0.202,
        "cacheRead": 0.123625
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-03-10",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.5-9B"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 139.32,
        "accessibility": "Open weights (restricted use)",
        "gpqa": 0.79,
        "gpqaStderr": 0.0237,
        "aime": 0.617,
        "aimeStderr": 0.06
      }
    },
    {
      "id": "qwen/qwen3.5-flash-02-23",
      "orId": "qwen/qwen3.5-flash-02-23",
      "vendor": "qwen",
      "model": "qwen3.5-flash-02-23",
      "name": "Qwen: Qwen3.5-Flash",
      "blurb": "The Qwen3.5 native vision-language Flash models are built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. Compared to the...",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "cost": {
        "input": 0.13,
        "output": 0.52
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-02-25",
      "bench": {
        "source": "epoch",
        "gpqa": 0.823,
        "gpqaStderr": 0.0272,
        "frontierMath": 0.062,
        "frontierMathStderr": 0.0142,
        "aime": 0.844,
        "aimeStderr": 0.0546,
        "simpleQA": 0.203,
        "simpleQAStderr": 0.0127
      }
    },
    {
      "id": "qwen/qwen3.5-plus-02-15",
      "orId": "qwen/qwen3.5-plus-02-15",
      "vendor": "qwen",
      "model": "qwen3.5-plus-02-15",
      "name": "Qwen: Qwen3.5 Plus 2026-02-15",
      "blurb": "The Qwen3.5 native vision-language series Plus models are built on a hybrid architecture that integrates linear attention mechanisms with sparse mixture-of-experts models, achieving higher inference efficiency. In a variety of...",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "cost": {
        "input": 0.52,
        "output": 3.12,
        "tiers": [
          {
            "minPromptTokens": 256000,
            "input": 0.65,
            "output": 3.9
          }
        ]
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-02-16"
    },
    {
      "id": "qwen/qwen3.5-plus-20260420",
      "orId": "qwen/qwen3.5-plus-20260420",
      "vendor": "qwen",
      "model": "qwen3.5-plus-20260420",
      "name": "Qwen: Qwen3.5 Plus 2026-04-20",
      "blurb": "Qwen3.5 Plus (April 2026) is a large-scale multimodal language model from Alibaba. It accepts text, image, and video input and produces text output, with a 1M token context window. This...",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "cost": {
        "input": 0.6,
        "output": 3.6,
        "cacheWrite": 0.75,
        "tiers": [
          {
            "minPromptTokens": 256000,
            "input": 0.75,
            "output": 4.5
          }
        ]
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-27"
    },
    {
      "id": "qwen/qwen3.6-27b",
      "orId": "qwen/qwen3.6-27b",
      "vendor": "qwen",
      "model": "qwen3.6-27b",
      "name": "Qwen: Qwen3.6 27B",
      "blurb": "Qwen3.6 27B is a dense 27-billion-parameter language model from the Qwen Team at Alibaba, released in April 2026. It features hybrid multimodal capabilities — accepting text, image, and video inputs...",
      "contextWindow": 262144,
      "maxOutput": 81920,
      "cost": {
        "input": 0.64,
        "output": 6.4
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-27",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.6-27B"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 146.73,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.859,
        "gpqaStderr": 0.0248,
        "aime": 0.911,
        "aimeStderr": 0.0429
      }
    },
    {
      "id": "qwen/qwen3.6-35b-a3b",
      "orId": "qwen/qwen3.6-35b-a3b",
      "vendor": "qwen",
      "model": "qwen3.6-35b-a3b",
      "name": "Qwen: Qwen3.6 35B A3B",
      "blurb": "Qwen3.6-35B-A3B is an open-weight multimodal model from Alibaba Cloud with 35 billion total parameters and 3 billion active parameters per token. It uses a hybrid sparse mixture-of-experts architecture combining Gated...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.2,
        "output": 1.8,
        "cacheRead": 0.05
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-27",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.6-35B-A3B"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 144.02,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.848,
        "gpqaStderr": 0.0255,
        "aime": 0.867,
        "aimeStderr": 0.0512
      }
    },
    {
      "id": "qwen/qwen3.6-flash",
      "orId": "qwen/qwen3.6-flash",
      "vendor": "qwen",
      "model": "qwen3.6-flash",
      "name": "Qwen: Qwen3.6 Flash",
      "blurb": "Qwen3.6 Flash is a fast, efficient language model from Alibaba's Qwen 3.6 series. It supports text, image, and video input with a 1M token context window. Tiered pricing kicks in...",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "cost": {
        "input": 0.375,
        "output": 2.25,
        "cacheWrite": 0.46875,
        "tiers": [
          {
            "minPromptTokens": 256000,
            "input": 1.5,
            "output": 6
          }
        ]
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-27",
      "bench": {
        "source": "epoch",
        "eci": 143.65,
        "accessibility": "Unknown",
        "gpqa": 0.833,
        "gpqaStderr": 0.0266,
        "frontierMath": 0.103,
        "frontierMathStderr": 0.0179,
        "aime": 0.844,
        "aimeStderr": 0.0546,
        "simpleQA": 0.159,
        "simpleQAStderr": 0.0116
      }
    },
    {
      "id": "qwen/qwen3.6-plus",
      "orId": "qwen/qwen3.6-plus",
      "vendor": "qwen",
      "model": "qwen3.6-plus",
      "name": "Qwen: Qwen3.6 Plus",
      "blurb": "Qwen 3.6 Plus builds on a hybrid architecture that combines efficient linear attention with sparse mixture-of-experts routing, enabling strong scalability and high-performance inference. Compared to the 3.5 series, it delivers...",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "cost": {
        "input": 0.65,
        "output": 3.9,
        "cacheWrite": 0.8125,
        "tiers": [
          {
            "minPromptTokens": 256000,
            "input": 2.6,
            "output": 7.8
          }
        ]
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-02",
      "bench": {
        "source": "epoch",
        "eci": 148.88,
        "accessibility": "API access",
        "gpqa": 0.884,
        "gpqaStderr": 0.0228,
        "sweBenchVerified": 0.579,
        "sweBenchVerifiedStderr": 0.0225,
        "frontierMath": 0.262,
        "frontierMathStderr": 0.0259,
        "aime": 0.933,
        "aimeStderr": 0.0376,
        "simpleQA": 0.441,
        "simpleQAStderr": 0.0157
      }
    },
    {
      "id": "qwen/qwen3.7-flash",
      "orId": "qwen/qwen3.7-flash",
      "vendor": "qwen",
      "model": "qwen3.7-flash",
      "name": "Qwen: Qwen3.7 Flash",
      "blurb": "Qwen3.7 Flash is a vision-language reasoning model from Alibaba. It is suited for multimodal agents, visual coding, search, and computer interaction, with strengths in object recognition, spatial understanding, and real-world...",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "cost": {
        "input": 0.06,
        "output": 0.26,
        "cacheRead": 0.012,
        "cacheWrite": 0.076,
        "tiers": [
          {
            "minPromptTokens": 32000,
            "input": 0.2,
            "output": 0.8,
            "cacheRead": 0.04
          },
          {
            "minPromptTokens": 256000,
            "input": 0.4,
            "output": 1.6,
            "cacheRead": 0.08
          }
        ]
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-07-27",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "gpqa": 0.823,
        "gpqaStderr": 0.0272,
        "aime": 0.867,
        "aimeStderr": 0.0512
      }
    },
    {
      "id": "qwen/qwen3.7-max",
      "orId": "qwen/qwen3.7-max",
      "vendor": "qwen",
      "model": "qwen3.7-max",
      "name": "Qwen: Qwen3.7 Max",
      "blurb": "Qwen3.7-Max is the flagship model in Alibaba's Qwen3.7 series. It supports text input and output and is designed for agent-centric workloads, with particular strengths in coding, office and productivity tasks,...",
      "contextWindow": 1000000,
      "maxOutput": 131072,
      "cost": {
        "input": 2.95,
        "output": 8.85,
        "cacheRead": 0.59,
        "cacheWrite": 3.6875
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-05-21",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 154.54,
        "accessibility": "API access",
        "gpqa": 0.909,
        "gpqaStderr": 0.0205,
        "sweBenchVerified": 0.773,
        "sweBenchVerifiedStderr": 0.0191,
        "aime": 0.956,
        "aimeStderr": 0.0311,
        "simpleQA": 0.558,
        "simpleQAStderr": 0.0157
      }
    },
    {
      "id": "qwen/qwen3.7-plus",
      "orId": "qwen/qwen3.7-plus",
      "vendor": "qwen",
      "model": "qwen3.7-plus",
      "name": "Qwen: Qwen3.7 Plus",
      "blurb": "Qwen3.7-Plus is a cost-effective model in Alibaba's Qwen3.7 series. It supports text and image input with text output, building on the series' text capabilities with a comprehensive upgrade to its...",
      "contextWindow": 1000000,
      "maxOutput": 131072,
      "cost": {
        "input": 0.64,
        "output": 2.56,
        "cacheRead": 0.128,
        "cacheWrite": 0.8,
        "tiers": [
          {
            "minPromptTokens": 256000,
            "input": 1.92,
            "output": 7.68,
            "cacheRead": 0.384
          }
        ]
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-06-03",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 147.87,
        "accessibility": "API access",
        "gpqa": 0.879,
        "gpqaStderr": 0.0233,
        "aime": 0.933,
        "aimeStderr": 0.0376
      }
    },
    {
      "id": "qwen/qwen3.7-plus:batch",
      "orId": "qwen/qwen3.7-plus:batch",
      "vendor": "qwen",
      "model": "qwen3.7-plus:batch",
      "name": "Qwen: Qwen3.7 Plus (batch)",
      "blurb": "Qwen3.7-Plus is a cost-effective model in Alibaba's Qwen3.7 series. It supports text and image input with text output, building on the series' text capabilities with a comprehensive upgrade to its...",
      "contextWindow": 262144,
      "maxOutput": 131072,
      "cost": {
        "input": 0.46,
        "output": 1.84
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-06-03"
    },
    {
      "id": "qwen/qwen3.8-2.4t-a95b",
      "orId": "qwen/qwen3.8-2.4t-a95b",
      "vendor": "qwen",
      "model": "qwen3.8-2.4t-a95b",
      "name": "Qwen: Qwen3.8 2.4T A95B",
      "blurb": "Qwen3.8 2.4T A95B is an open-weight sparse mixture-of-experts model from Qwen and the open-weight variant of [Qwen3.8 Max](/qwen/qwen3.8-max), with 95 billion active parameters out of 2.4 trillion total. It is...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 4,
        "output": 12,
        "cacheRead": 0.5
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "medium",
          "low"
        ],
        "defaultEffort": "xhigh"
      },
      "tools": true,
      "released": "2026-08-12",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.8-2.4T-A95B"
      }
    },
    {
      "id": "qwen/qwen3.8-27b",
      "orId": "qwen/qwen3.8-27b",
      "vendor": "qwen",
      "model": "qwen3.8-27b",
      "name": "Qwen: Qwen3.8 27B",
      "blurb": "Qwen3.8 27B is an open-weight dense vision-language model from Qwen. It is suited for coding, professional workflows, research, multimodal interaction, and long-running agent tasks, with flexible thinking that can be...",
      "contextWindow": 1024000,
      "maxOutput": 235929,
      "cost": {
        "input": 0.3,
        "output": 3.75,
        "cacheRead": 0.075
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "xhigh"
      },
      "tools": true,
      "released": "2026-08-14",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.8-27B"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 152.64,
        "accessibility": "Open weights (unrestricted)"
      }
    },
    {
      "id": "qwen/qwen3.8-27b:batch",
      "orId": "qwen/qwen3.8-27b:batch",
      "vendor": "qwen",
      "model": "qwen3.8-27b:batch",
      "name": "Qwen: Qwen3.8 27B (batch)",
      "blurb": "Qwen3.8 27B is an open-weight dense vision-language model from Qwen. It is suited for coding, professional workflows, research, multimodal interaction, and long-running agent tasks, with flexible thinking that can be...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.32,
        "output": 4,
        "cacheRead": 0.08
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "xhigh"
      },
      "tools": true,
      "released": "2026-08-14",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.8-27B"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "qwen/qwen3.8-flash",
      "orId": "qwen/qwen3.8-flash",
      "vendor": "qwen",
      "model": "qwen3.8-flash",
      "name": "Qwen: Qwen3.8 Flash",
      "blurb": "Qwen3.8 Flash is a multimodal reasoning model from Alibaba. It is suited for coding assistance, agentic workflows, visual understanding, document and codebase analysis, desktop interaction, chart analysis, and long-video analysis.",
      "contextWindow": 1000000,
      "maxOutput": 131072,
      "cost": {
        "input": 0.3,
        "output": 0.94,
        "cacheRead": 0.032
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-08-26",
      "openWeights": {
        "huggingFace": "Qwen/Qwen3.8-Flash-Next"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "qwen/qwen3.8-max-0902",
      "orId": "qwen/qwen3.8-max-0902",
      "vendor": "qwen",
      "model": "qwen3.8-max-0902",
      "name": "Qwen: Qwen3.8 Max (0902)",
      "blurb": "Qwen3.8 Max 0902 is an updated snapshot of Qwen3.8 Max from Alibaba's Qwen team. It is a 2.4-trillion-parameter mixture-of-experts model that accepts text, image, and video input and returns text,...",
      "contextWindow": 1000000,
      "maxOutput": 131072,
      "cost": {
        "input": 4,
        "output": 12,
        "cacheRead": 0.5
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "xhigh"
      },
      "tools": true,
      "released": "2026-09-03",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 155.97,
        "accessibility": "API access",
        "gpqa": 0.923,
        "gpqaStderr": 0.0172,
        "aime": 1,
        "aimeStderr": 0,
        "simpleQA": 0.473,
        "simpleQAStderr": 0.0158
      }
    },
    {
      "id": "qwen/qwen3.8-max-0902:batch",
      "orId": "qwen/qwen3.8-max-0902:batch",
      "vendor": "qwen",
      "model": "qwen3.8-max-0902:batch",
      "name": "Qwen: Qwen3.8 Max (0902) (batch)",
      "blurb": "Qwen3.8 Max 0902 is an updated snapshot of Qwen3.8 Max from Alibaba's Qwen team. It is a 2.4-trillion-parameter mixture-of-experts model that accepts text, image, and video input and returns text,...",
      "contextWindow": 262144,
      "maxOutput": 131072,
      "cost": {
        "input": 2.3,
        "output": 6.9
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "xhigh"
      },
      "tools": true,
      "released": "2026-09-03"
    },
    {
      "id": "qwen/qwen3.8-max-prime",
      "orId": "qwen/qwen3.8-max-prime",
      "vendor": "qwen",
      "model": "qwen3.8-max-prime",
      "name": "Qwen: Qwen3.8 Max Prime",
      "blurb": "Qwen3.8 Max Prime is a higher-throughput variant of Qwen3.8 Max from Alibaba's Qwen team, served as a separate SKU at a higher price point. It accepts text, image, and video...",
      "contextWindow": 1000000,
      "maxOutput": 131072,
      "cost": {
        "input": 8,
        "output": 24,
        "cacheRead": 1
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "xhigh"
      },
      "tools": true,
      "released": "2026-09-23"
    },
    {
      "id": "qwen/qwen3.8-omni-flash",
      "orId": "qwen/qwen3.8-omni-flash",
      "vendor": "qwen",
      "model": "qwen3.8-omni-flash",
      "name": "Qwen: Qwen3.8 Omni Flash",
      "blurb": "Qwen3.8 Omni Flash is an omni-modal reasoning model from Alibaba, the first Qwen model built around agentic capabilities with native audio-video understanding. It is suited for audio-video analysis and summarization,...",
      "contextWindow": 1000000,
      "maxOutput": 131072,
      "cost": {
        "input": 0.3,
        "output": 0.94,
        "cacheRead": 0.032
      },
      "input": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-09-21"
    },
    {
      "id": "qwen/qwen4-max",
      "orId": "qwen/qwen4-max",
      "vendor": "qwen",
      "model": "qwen4-max",
      "name": "Qwen: Qwen4 Max",
      "blurb": "Qwen4 Max is the flagship model in Alibaba's Qwen4 series. It accepts text, image, and video input and returns text, with a 1M-token context window and reasoning enabled by default....",
      "contextWindow": 1000000,
      "maxOutput": 393216,
      "cost": {
        "input": 3,
        "output": 9,
        "cacheRead": 0.15,
        "cacheWrite": 3.75
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-11-16"
    },
    {
      "id": "recursive/flint-v1.0:free",
      "orId": "recursive/flint-v1.0:free",
      "vendor": "recursive",
      "model": "flint-v1.0:free",
      "name": "Recursive: Flint (free)",
      "blurb": "Flint is an open-weight mixture-of-experts model from Recursive, with about 2.5B active parameters out of 40B total, built by Recursive's autoresearch system Eureka Machine for coding agents. It is suited...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-10-21",
      "openWeights": {
        "huggingFace": "RecursiveSI/flint"
      }
    },
    {
      "id": "reflection/beam-1.0-501b-a23b",
      "orId": "reflection/beam-1.0-501b-a23b",
      "vendor": "reflection",
      "model": "beam-1.0-501b-a23b",
      "name": "Reflection: Beam 1.0 501B A23B",
      "blurb": "Beam 1.0 is Reflection's open-weight reasoning model for coding and long-horizon agentic workflows. It uses a sparse mixture-of-experts architecture with 501B total and 23B active parameters per token, and a...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.6,
        "output": 2,
        "cacheRead": 0.08
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "knowledgeCutoff": "2026-06-30",
      "released": "2026-10-16",
      "openWeights": {
        "huggingFace": "reflection/Beam-1.0-501B-A23B"
      }
    },
    {
      "id": "rekaai/reka-edge",
      "orId": "rekaai/reka-edge",
      "vendor": "rekaai",
      "model": "reka-edge",
      "name": "Reka Edge",
      "blurb": "Reka Edge is an extremely efficient 7B multimodal vision-language model that accepts image/video+text inputs and generates text outputs. This model is optimized specifically to deliver industry-leading performance in image understanding,...",
      "contextWindow": 16384,
      "maxOutput": 14745,
      "cost": {
        "input": 0.2,
        "output": 0.2
      },
      "input": [
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-03-20",
      "openWeights": {
        "huggingFace": "RekaAI/reka-edge-2603"
      }
    },
    {
      "id": "rekaai/reka-flash-3",
      "orId": "rekaai/reka-flash-3",
      "vendor": "rekaai",
      "model": "reka-flash-3",
      "name": "Reka Flash 3",
      "blurb": "Reka Flash 3 is a general-purpose, instruction-tuned large language model with 21 billion parameters, developed by Reka. It excels at general chat, coding tasks, instruction-following, and function calling. Featuring a...",
      "contextWindow": 65536,
      "maxOutput": 58982,
      "cost": {
        "input": 0.2,
        "output": 0.4
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": false,
      "knowledgeCutoff": "2025-01-31",
      "released": "2025-03-12",
      "openWeights": {
        "huggingFace": "RekaAI/reka-flash-3"
      }
    },
    {
      "id": "relace/relace-apply-3",
      "orId": "relace/relace-apply-3",
      "vendor": "relace",
      "model": "relace-apply-3",
      "name": "Relace: Relace Apply 3",
      "blurb": "Relace Apply 3 is a specialized code-patching LLM that merges AI-suggested edits straight into your source files. It can apply updates from GPT-4o, Claude, and others into your files at...",
      "contextWindow": 256000,
      "maxOutput": 128000,
      "cost": {
        "input": 1.7,
        "output": 2.5
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2025-09-26"
    },
    {
      "id": "relace/relace-search",
      "orId": "relace/relace-search",
      "vendor": "relace",
      "model": "relace-search",
      "name": "Relace: Relace Search",
      "blurb": "The relace-search model uses 4-12 `view_file` and `grep` tools in parallel to explore a codebase and return relevant files to the user request. In contrast to RAG, relace-search performs agentic...",
      "contextWindow": 256000,
      "maxOutput": 128000,
      "cost": {
        "input": 2,
        "output": 6
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2025-12-08"
    },
    {
      "id": "ricursiveai/ripple-65b-preview",
      "orId": "ricursiveai/ripple-65b-preview",
      "vendor": "ricursiveai",
      "model": "ripple-65b-preview",
      "name": "Ricursive: Ripple 65B Preview",
      "blurb": "Ripple-65B-Preview by ricursiveAI, a 65B model with native tool calling, served in BF16.",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0.2,
        "output": 0.6,
        "cacheRead": 0.05
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-11-04",
      "openWeights": {
        "huggingFace": "ricursiveAI/Ripple-65B-Preview"
      }
    },
    {
      "id": "sakana/fugu-max",
      "orId": "sakana/fugu-max",
      "vendor": "sakana",
      "model": "fugu-max",
      "name": "Sakana: Fugu Max",
      "blurb": "Fugu Max is the cost-performance model in Sakana AI's Fugu family. Rather than a single monolithic model, Fugu is a learned multi-agent orchestration system: a language model trained to route...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 4,
        "output": 12,
        "cacheRead": 0.5,
        "webSearch": 0.02
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-11"
    },
    {
      "id": "sakana/fugu-ultra",
      "orId": "sakana/fugu-ultra",
      "vendor": "sakana",
      "model": "fugu-ultra",
      "name": "Sakana: Fugu Ultra",
      "blurb": "Fugu Ultra is the higher-performance model in Sakana AI's Fugu family. Rather than a single monolithic model, Fugu is a learned multi-agent orchestration system: a language model trained to route...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 60,
        "cacheRead": 1,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 20,
            "output": 90,
            "cacheRead": 2
          }
        ]
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high"
        ],
        "defaultEffort": "xhigh"
      },
      "tools": true,
      "released": "2026-06-24"
    },
    {
      "id": "sakana/fugu-ultra-v2",
      "orId": "sakana/fugu-ultra-v2",
      "vendor": "sakana",
      "model": "fugu-ultra-v2",
      "name": "Sakana: Fugu Ultra v2",
      "blurb": "Fugu Ultra v2 is the higher-performance model in Sakana AI's Fugu family. Rather than a single monolithic model, Fugu is a learned multi-agent orchestration system: a language model trained to...",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "cost": {
        "input": 10,
        "output": 60,
        "cacheRead": 1,
        "webSearch": 0.02,
        "tiers": [
          {
            "minPromptTokens": 272000,
            "input": 20,
            "output": 90,
            "cacheRead": 2
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "xhigh",
          "high"
        ],
        "defaultEffort": "xhigh"
      },
      "tools": true,
      "knowledgeCutoff": "2026-08-28",
      "released": "2026-09-11"
    },
    {
      "id": "sakana/sakana-namazu",
      "orId": "sakana/sakana-namazu",
      "vendor": "sakana",
      "model": "sakana-namazu",
      "name": "Sakana: Sakana Namazu",
      "blurb": "Sakana Namazu is a Japanese-specialized reasoning model from Sakana AI, based on Kimi K2.6 with additional training for Japanese language and business contexts. It is suited for Japanese instruction following,...",
      "contextWindow": 262144,
      "maxOutput": 65536,
      "cost": {
        "input": 1.9,
        "output": 8,
        "cacheRead": 0.3,
        "webSearch": 0.014
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-08-11"
    },
    {
      "id": "sao10k/l3-lunaris-8b",
      "orId": "sao10k/l3-lunaris-8b",
      "vendor": "sao10k",
      "model": "l3-lunaris-8b",
      "name": "Sao10K: Llama 3 8B Lunaris",
      "blurb": "Lunaris 8B is a versatile generalist and roleplaying model based on Llama 3. It's a strategic merge of multiple models, designed to balance creativity with improved logic and general knowledge....",
      "contextWindow": 8192,
      "maxOutput": 7372,
      "cost": {
        "input": 0.08,
        "output": 0.1
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2023-12-31",
      "released": "2024-08-13",
      "openWeights": {
        "huggingFace": "Sao10K/L3-8B-Lunaris-v1"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "sao10k/l3.1-euryale-70b",
      "orId": "sao10k/l3.1-euryale-70b",
      "vendor": "sao10k",
      "model": "l3.1-euryale-70b",
      "name": "Sao10K: Llama 3.1 Euryale 70B v2.2",
      "blurb": "Euryale L3.1 70B v2.2 is a model focused on creative roleplay from [Sao10k](https://ko-fi.com/sao10k). It is the successor of [Euryale L3 70B v2.1](/models/sao10k/l3-euryale-70b).",
      "contextWindow": 131072,
      "maxOutput": 16384,
      "cost": {
        "input": 1.7,
        "output": 1.7
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2023-12-31",
      "released": "2024-08-28",
      "openWeights": {
        "huggingFace": "Sao10K/L3.1-70B-Euryale-v2.2"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "scaledown/extract",
      "orId": "scaledown/extract",
      "vendor": "scaledown",
      "model": "extract",
      "name": "ScaleDown: Extract",
      "blurb": "Extract is a description-driven entity extraction model from ScaleDown. Instead of a fixed set of entity types, callers describe the fields they want in plain English through `provider.options.scaledown.entities`, and the...",
      "contextWindow": 1000000,
      "maxOutput": 900000,
      "cost": {
        "input": 0.08,
        "output": 0
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2026-10-13"
    },
    {
      "id": "scaledown/summarize",
      "orId": "scaledown/summarize",
      "vendor": "scaledown",
      "model": "summarize",
      "name": "ScaleDown: Summarize",
      "blurb": "Summarize is an abstractive summarization model from ScaleDown. It condenses long text and documents into a fluent, faithful rewrite rather than lifting sentences from the source, and is suited for...",
      "contextWindow": 1000000,
      "maxOutput": 4000,
      "cost": {
        "input": 0.08,
        "output": 0
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2026-10-13"
    },
    {
      "id": "stepfun/step-3.7-flash",
      "orId": "stepfun/step-3.7-flash",
      "vendor": "stepfun",
      "model": "step-3.7-flash",
      "name": "StepFun: Step 3.7 Flash",
      "blurb": "Step 3.7 Flash is StepFun's latest high-efficiency multimodal Mixture-of-Experts model. It pairs a 196B-parameter language backbone with a vision encoder for native image and video understanding, activating roughly 11B parameters...",
      "contextWindow": 262144,
      "maxOutput": 230400,
      "cost": {
        "input": 0.4,
        "output": 2.3,
        "cacheRead": 0.08
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-05-28",
      "openWeights": {
        "huggingFace": "stepfun-ai/Step-3.7-Flash"
      }
    },
    {
      "id": "stepfun/step-5-preview",
      "orId": "stepfun/step-5-preview",
      "vendor": "stepfun",
      "model": "step-5-preview",
      "name": "StepFun: Step 5 Preview",
      "blurb": "Step 5 Preview is StepFun's flagship model for agentic work, built on a sparse Mixture-of-Experts architecture (27B active / 600B total parameters). It performs strongly in software engineering and professional...",
      "contextWindow": 1000000,
      "maxOutput": 64000,
      "cost": {
        "input": 2,
        "output": 5.4,
        "cacheRead": 0.2
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-10-08",
      "openWeights": {
        "huggingFace": "stepfun-ai/Step-5-Preview"
      }
    },
    {
      "id": "tencent/hy-mt2-1.8b",
      "orId": "tencent/hy-mt2-1.8b",
      "vendor": "tencent",
      "model": "hy-mt2-1.8b",
      "name": "Tencent: Hy-MT2-1.8B",
      "blurb": "Hy-MT2-1.8B is a compact 1.8B-parameter translation model from Tencent. It supports 33 language pairs and five Chinese dialect and minority-language pairs, with workflows for structured, delimiter-based, contextual, glossary-based, and style-guided...",
      "contextWindow": 8192,
      "maxOutput": 4096,
      "cost": {
        "input": 0.088,
        "output": 0.354
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2026-08-20",
      "openWeights": {
        "huggingFace": "tencent/Hy-MT2-1.8B"
      }
    },
    {
      "id": "tencent/hy-mt2-30b-a3b",
      "orId": "tencent/hy-mt2-30b-a3b",
      "vendor": "tencent",
      "model": "hy-mt2-30b-a3b",
      "name": "Tencent: Hy-MT2-30B-A3B",
      "blurb": "Hy-MT2-30B-A3B is Tencent's flagship translation model in the Hy-MT2 family. It supports 33 language pairs and five Chinese dialect and minority-language pairs, with workflows for structured, delimiter-based, contextual, glossary-based, and...",
      "contextWindow": 8192,
      "maxOutput": 4096,
      "cost": {
        "input": 0.148,
        "output": 0.59
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2026-08-20",
      "openWeights": {
        "huggingFace": "tencent/Hy-MT2-30B-A3B"
      }
    },
    {
      "id": "tencent/hy-mt2-7b",
      "orId": "tencent/hy-mt2-7b",
      "vendor": "tencent",
      "model": "hy-mt2-7b",
      "name": "Tencent: Hy-MT2-7B",
      "blurb": "Hy-MT2-7B is a 7B-parameter translation model from Tencent. It supports 33 language pairs and five Chinese dialect and minority-language pairs, with workflows for structured, delimiter-based, contextual, glossary-based, and style-guided translation.",
      "contextWindow": 8192,
      "maxOutput": 4096,
      "cost": {
        "input": 0.148,
        "output": 0.59
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2026-08-19",
      "openWeights": {
        "huggingFace": "tencent/Hy-MT2-7B"
      }
    },
    {
      "id": "tencent/hy3",
      "orId": "tencent/hy3",
      "vendor": "tencent",
      "model": "hy3",
      "name": "Tencent: Hy3",
      "blurb": "Hy3 is a 295B-parameter Mixture-of-Experts model from Tencent (21B active, 192 experts with top-8 routing) built for reasoning, agentic workflows, and real-world production use. It supports a configurable reasoning effort:...",
      "contextWindow": 262144,
      "maxOutput": 128000,
      "cost": {
        "input": 0.264,
        "output": 1.056,
        "cacheRead": 0.066,
        "tiers": [
          {
            "input": 0.264,
            "output": 1.056,
            "cacheRead": 0.066
          },
          {
            "input": 0.165,
            "output": 0.66,
            "cacheRead": 0.04125
          }
        ]
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-07-06",
      "openWeights": {
        "huggingFace": "tencent/Hy3"
      }
    },
    {
      "id": "tencent/hy3-preview",
      "orId": "tencent/hy3-preview",
      "vendor": "tencent",
      "model": "hy3-preview",
      "name": "Tencent: Hy3 preview",
      "blurb": "Hy3 preview is a high-efficiency Mixture-of-Experts model from Tencent designed for agentic workflows and production use. It supports configurable reasoning levels across disabled, low, and high modes, allowing it to...",
      "contextWindow": 262144,
      "maxOutput": 235929,
      "cost": {
        "input": 0.36,
        "output": 1.2,
        "cacheRead": 0.12
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-04-22",
      "openWeights": {
        "huggingFace": "tencent/Hy3-preview"
      }
    },
    {
      "id": "tencent/hy4",
      "orId": "tencent/hy4",
      "vendor": "tencent",
      "model": "hy4",
      "name": "Tencent: Hy4",
      "blurb": "The official release of Hy4 delivers comprehensive and substantial improvements over Hy4 Preview, with particularly strong gains in agentic capabilities and performance across a wide range of long-horizon, real-world productivity...",
      "contextWindow": 1048576,
      "maxOutput": 65536,
      "cost": {
        "input": 0.834,
        "output": 2.501,
        "cacheRead": 0.042
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-11-04",
      "openWeights": {
        "huggingFace": "tencent/Hy4"
      }
    },
    {
      "id": "tencent/hy4-preview",
      "orId": "tencent/hy4-preview",
      "vendor": "tencent",
      "model": "hy4-preview",
      "name": "Tencent: Hy4 preview",
      "blurb": "Tencent: Hy4 preview is a mixture-of-experts model from Tencent, with 49B active parameters out of 770B total. It is designed for coding agents, complex tool-use workflows, and productivity tasks that...",
      "contextWindow": 1048576,
      "maxOutput": 64000,
      "cost": {
        "input": 1.668,
        "output": 5.002,
        "cacheRead": 0.084,
        "tiers": [
          {
            "input": 1.668,
            "output": 5.002,
            "cacheRead": 0.084
          },
          {
            "input": 1.5012,
            "output": 4.5018,
            "cacheRead": 0.0756
          }
        ]
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "low",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-08-28",
      "openWeights": {
        "huggingFace": "tencent/Hy4-preview"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "thedrummer/cydonia-24b-v4.1",
      "orId": "thedrummer/cydonia-24b-v4.1",
      "vendor": "thedrummer",
      "model": "cydonia-24b-v4.1",
      "name": "TheDrummer: Cydonia 24B V4.1",
      "blurb": "Uncensored and creative writing model based on Mistral Small 3.2 24B with good recall, prompt adherence, and intelligence.",
      "contextWindow": 131072,
      "maxOutput": 117964,
      "cost": {
        "input": 0.6,
        "output": 1,
        "cacheRead": 0.3
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-04-30",
      "released": "2025-09-27",
      "openWeights": {
        "huggingFace": "thedrummer/cydonia-24b-v4.1"
      }
    },
    {
      "id": "thedrummer/skyfall-36b-v2",
      "orId": "thedrummer/skyfall-36b-v2",
      "vendor": "thedrummer",
      "model": "skyfall-36b-v2",
      "name": "TheDrummer: Skyfall 36B V2",
      "blurb": "Skyfall 36B v2 is an enhanced iteration of Mistral Small 2501, specifically fine-tuned for improved creativity, nuanced writing, role-playing, and coherent storytelling.",
      "contextWindow": 32768,
      "maxOutput": 29491,
      "cost": {
        "input": 1.1,
        "output": 1.6,
        "cacheRead": 0.5
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-06-30",
      "released": "2025-03-10",
      "openWeights": {
        "huggingFace": "TheDrummer/Skyfall-36B-v2"
      }
    },
    {
      "id": "thedrummer/unslopnemo-12b",
      "orId": "thedrummer/unslopnemo-12b",
      "vendor": "thedrummer",
      "model": "unslopnemo-12b",
      "name": "TheDrummer: UnslopNemo 12B",
      "blurb": "UnslopNemo v4.1 is the latest addition from the creator of Rocinante, designed for adventure writing and role-play scenarios.",
      "contextWindow": 1024000,
      "maxOutput": 819200,
      "cost": {
        "input": 0.8,
        "output": 0.8
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2024-04-30",
      "released": "2024-11-08",
      "openWeights": {
        "huggingFace": "TheDrummer/UnslopNemo-12B-v4.1"
      }
    },
    {
      "id": "thinkingmachines/inkling",
      "orId": "thinkingmachines/inkling",
      "vendor": "thinkingmachines",
      "model": "inkling",
      "name": "Thinking Machines: Inkling",
      "blurb": "Inkling is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 41B active parameters out of 975B total. It is designed for general-purpose reasoning, coding, agentic and tool-use systems,...",
      "contextWindow": 524288,
      "maxOutput": 471859,
      "cost": {
        "input": 2,
        "output": 8.1,
        "cacheRead": 0.34
      },
      "input": [
        "text",
        "image",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-07-17",
      "openWeights": {
        "huggingFace": "thinkingmachines/Inkling"
      },
      "bench": {
        "source": "epoch",
        "eci": 148.39,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.883,
        "gpqaStderr": 0.0194,
        "aime": 0.889,
        "aimeStderr": 0.0392,
        "simpleQA": 0.403,
        "simpleQAStderr": 0.0155
      }
    },
    {
      "id": "thinkingmachines/inkling-2-small",
      "orId": "thinkingmachines/inkling-2-small",
      "vendor": "thinkingmachines",
      "model": "inkling-2-small",
      "name": "Thinking Machines: Inkling 2 Small",
      "blurb": "Inkling 2 Small is an open-weight multimodal mixture-of-experts reasoning model from Thinking Machines Lab, with 17B active parameters out of 594B total. It is the first model in the second...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.6,
        "output": 2,
        "cacheRead": 0.06
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-11-09",
      "openWeights": {
        "huggingFace": "thinkingmachines/Inkling-2-Small"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "thinkingmachines/inkling-2-small:free",
      "orId": "thinkingmachines/inkling-2-small:free",
      "vendor": "thinkingmachines",
      "model": "inkling-2-small:free",
      "name": "Thinking Machines: Inkling 2 Small (free)",
      "blurb": "Inkling 2 Small is an open-weight multimodal mixture-of-experts reasoning model from Thinking Machines Lab, with 17B active parameters out of 594B total. It is the first model in the second...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0,
        "output": 0
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "medium",
          "low",
          "minimal"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-11-09",
      "openWeights": {
        "huggingFace": "thinkingmachines/Inkling-2-Small"
      }
    },
    {
      "id": "thinkingmachines/inkling-small",
      "orId": "thinkingmachines/inkling-small",
      "vendor": "thinkingmachines",
      "model": "inkling-small",
      "name": "Thinking Machines: Inkling Small",
      "blurb": "Inkling Small is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 12B active parameters out of 276B total. It is positioned as the smaller, more efficient member of...",
      "contextWindow": 524288,
      "maxOutput": 262144,
      "cost": {
        "input": 0.9,
        "output": 2.4,
        "cacheRead": 0.2
      },
      "input": [
        "text",
        "image",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-07-30",
      "openWeights": {
        "huggingFace": "thinkingmachines/Inkling-Small"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 150.75,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.885,
        "gpqaStderr": 0.0191,
        "aime": 0.9,
        "aimeStderr": 0.0396,
        "simpleQA": 0.191,
        "simpleQAStderr": 0.0124
      }
    },
    {
      "id": "thinkingmachines/inkling-small:batch",
      "orId": "thinkingmachines/inkling-small:batch",
      "vendor": "thinkingmachines",
      "model": "inkling-small:batch",
      "name": "Thinking Machines: Inkling Small (batch)",
      "blurb": "Inkling Small is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 12B active parameters out of 276B total. It is positioned as the smaller, more efficient member of...",
      "contextWindow": 524288,
      "maxOutput": 262144,
      "cost": {
        "input": 0.72,
        "output": 1.92,
        "cacheRead": 0.16
      },
      "input": [
        "text",
        "image",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-07-30",
      "openWeights": {
        "huggingFace": "thinkingmachines/Inkling-Small"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "typesafe/jev-router",
      "orId": "typesafe/jev-router",
      "vendor": "typesafe",
      "model": "jev-router",
      "name": "TypeSafe: Jev Router",
      "blurb": "Jev Router picks the best model and reasoning effort for each request, balancing quality, speed, and cost. It runs on Jev, TypeSafe's first System One model, and adapts as your...",
      "contextWindow": 1000000,
      "maxOutput": 64000,
      "input": [
        "audio",
        "file",
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2026-09-25"
    },
    {
      "id": "unbiased/pareto-26.10",
      "orId": "unbiased/pareto-26.10",
      "vendor": "unbiased",
      "model": "pareto-26.10",
      "name": "Pareto 26.10",
      "blurb": "Pareto is a multimodal composite model built for research, coding, and agentic workflows, while delivering frontier-level performance across a broad range of general-purpose tasks.",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 1.6,
        "output": 6.4,
        "cacheRead": 0.06
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2026-10-15"
    },
    {
      "id": "unbiased/pareto-26.10-preview",
      "orId": "unbiased/pareto-26.10-preview",
      "vendor": "unbiased",
      "model": "pareto-26.10-preview",
      "name": "Pareto 26.10 Preview",
      "blurb": "Pareto is a multimodal composite model built for research, coding, and agentic workflows, while delivering frontier-level performance across a broad range of general-purpose tasks. This is a preview of the...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 1.6,
        "output": 6.4,
        "cacheRead": 0.06
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2026-10-01"
    },
    {
      "id": "unbiased/pareto-26.9",
      "orId": "unbiased/pareto-26.9",
      "vendor": "unbiased",
      "model": "pareto-26.9",
      "name": "Pareto 26.9",
      "blurb": "Pareto is a multimodal composite model built for research, coding, and agentic workflows, while delivering frontier-level performance across a broad range of general-purpose tasks.",
      "contextWindow": 262144,
      "maxOutput": 131072,
      "cost": {
        "input": 5,
        "output": 15,
        "cacheRead": 0.5
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "tools": true,
      "released": "2026-10-13"
    },
    {
      "id": "undi95/remm-slerp-l2-13b",
      "orId": "undi95/remm-slerp-l2-13b",
      "vendor": "undi95",
      "model": "remm-slerp-l2-13b",
      "name": "ReMM SLERP 13B",
      "blurb": "A recreation trial of the original MythoMax-L2-B13 but with updated models. #merge",
      "contextWindow": 6144,
      "maxOutput": 5529,
      "cost": {
        "input": 0.6,
        "output": 1,
        "cacheRead": 0.4
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "knowledgeCutoff": "2023-06-30",
      "released": "2023-07-22",
      "openWeights": {
        "huggingFace": "Undi95/ReMM-SLERP-L2-13B"
      }
    },
    {
      "id": "upstage/solar-mini4",
      "orId": "upstage/solar-mini4",
      "vendor": "upstage",
      "model": "solar-mini4",
      "name": "Upstage: Solar Mini 4",
      "blurb": "Solar Mini 4 is Upstage's compact, cost-efficient language model, a 35B-parameter mixture-of-experts with 3B active parameters and a 524K context window. It is built for agentic use cases where response...",
      "contextWindow": 524288,
      "maxOutput": 131072,
      "cost": {
        "input": 0.1,
        "output": 0.4,
        "cacheRead": 0.01
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-09-23"
    },
    {
      "id": "upstage/solar-pro4",
      "orId": "upstage/solar-pro4",
      "vendor": "upstage",
      "model": "solar-pro4",
      "name": "Upstage: Solar Pro 4",
      "blurb": "Solar Pro 4 is Upstage's cost-efficient large language model, featuring a 524K context window. It is built for long-horizon tasks and agentic workflows, with strong capabilities in office productivity, document-intensive...",
      "contextWindow": 524288,
      "maxOutput": 131072,
      "cost": {
        "input": 0.3,
        "output": 1.2,
        "cacheRead": 0.06
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "max",
          "xhigh",
          "high",
          "medium",
          "low",
          "minimal",
          "none"
        ],
        "defaultEffort": "medium"
      },
      "tools": true,
      "released": "2026-08-10"
    },
    {
      "id": "writer/palmyra-x5",
      "orId": "writer/palmyra-x5",
      "vendor": "writer",
      "model": "palmyra-x5",
      "name": "Writer: Palmyra X5",
      "blurb": "Palmyra X5 is Writer's most advanced model, purpose-built for building and scaling AI agents across the enterprise. It delivers industry-leading speed and efficiency on context windows up to 1 million...",
      "contextWindow": 1040000,
      "maxOutput": 8192,
      "cost": {
        "input": 1.2,
        "output": 12
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "tools": false,
      "released": "2026-01-21"
    },
    {
      "id": "x-ai/grok-4.20",
      "orId": "x-ai/grok-4.20",
      "vendor": "x-ai",
      "model": "grok-4.20",
      "name": "SpaceXAI: Grok 4.20",
      "blurb": "Grok 4.20 is a reasoning model from SpaceXAI with industry-leading speed and agentic tool calling capabilities. It combines the lowest hallucination rate on the market with strict prompt adherance, delivering...",
      "contextWindow": 2000000,
      "maxOutput": 1800000,
      "cost": {
        "input": 2.5,
        "output": 5,
        "cacheRead": 0.4,
        "webSearch": 0.01,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 5,
            "output": 10,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-09-01",
      "released": "2026-03-31",
      "bench": {
        "source": "epoch",
        "eci": 152.28,
        "accessibility": "API access"
      }
    },
    {
      "id": "x-ai/grok-4.20-multi-agent",
      "orId": "x-ai/grok-4.20-multi-agent",
      "vendor": "x-ai",
      "model": "grok-4.20-multi-agent",
      "name": "SpaceXAI: Grok 4.20 Multi-Agent",
      "blurb": "Grok 4.20 Multi-Agent is a variant of SpaceXAI’s Grok 4.20 designed for collaborative, agent-based workflows. Multiple agents operate in parallel to conduct deep research, coordinate tool use, and synthesize information...",
      "contextWindow": 2000000,
      "maxOutput": 1800000,
      "cost": {
        "input": 2.5,
        "output": 5,
        "cacheRead": 0.4,
        "webSearch": 0.01,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 5,
            "output": 10,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "medium"
      },
      "tools": false,
      "knowledgeCutoff": "2025-09-01",
      "released": "2026-03-31"
    },
    {
      "id": "x-ai/grok-4.3",
      "orId": "x-ai/grok-4.3",
      "vendor": "x-ai",
      "model": "grok-4.3",
      "name": "SpaceXAI: Grok 4.3",
      "blurb": "Grok 4.3 is a reasoning model from SpaceXAI. It accepts text and image inputs with text output, and is suited for agentic workflows, instruction-following tasks, and applications requiring high factual...",
      "contextWindow": 1000000,
      "maxOutput": 900000,
      "cost": {
        "input": 2.5,
        "output": 5,
        "cacheRead": 0.4,
        "webSearch": 0.01,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 5,
            "output": 10,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "low"
      },
      "tools": true,
      "released": "2026-04-30",
      "bench": {
        "source": "epoch",
        "gpqa": 0.888,
        "gpqaStderr": 0.0196,
        "aime": 0.933,
        "aimeStderr": 0.0304,
        "simpleQA": 0.332,
        "simpleQAStderr": 0.0149
      }
    },
    {
      "id": "x-ai/grok-4.3:batch",
      "orId": "x-ai/grok-4.3:batch",
      "vendor": "x-ai",
      "model": "grok-4.3:batch",
      "name": "SpaceXAI: Grok 4.3 (batch)",
      "blurb": "Grok 4.3 is a reasoning model from SpaceXAI. It accepts text and image inputs with text output, and is suited for agentic workflows, instruction-following tasks, and applications requiring high factual...",
      "contextWindow": 1000000,
      "maxOutput": 900000,
      "cost": {
        "input": 2,
        "output": 4,
        "cacheRead": 0.32,
        "webSearch": 0.01,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 4,
            "output": 8,
            "cacheRead": 0.64
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "high",
          "medium",
          "low",
          "none"
        ],
        "defaultEffort": "low"
      },
      "tools": true,
      "released": "2026-04-30"
    },
    {
      "id": "x-ai/grok-4.5",
      "orId": "x-ai/grok-4.5",
      "vendor": "x-ai",
      "model": "grok-4.5",
      "name": "SpaceXAI: Grok 4.5",
      "blurb": "Grok 4.5 is a model from SpaceXAI with frontier performance on coding, knowledge work, and STEM.",
      "contextWindow": 500000,
      "maxOutput": 450000,
      "cost": {
        "input": 4,
        "output": 12,
        "cacheRead": 0.6,
        "webSearch": 0.01,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 8,
            "output": 24,
            "cacheRead": 1.2
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-07-08",
      "bench": {
        "source": "epoch",
        "eci": 154.98,
        "accessibility": "API access",
        "gpqa": 0.934,
        "gpqaStderr": 0.0143,
        "aime": 0.978,
        "aimeStderr": 0.0127,
        "simpleQA": 0.483,
        "simpleQAStderr": 0.0158
      }
    },
    {
      "id": "x-ai/grok-4.6",
      "orId": "x-ai/grok-4.6",
      "vendor": "x-ai",
      "model": "grok-4.6",
      "name": "SpaceXAI: Grok 4.6",
      "blurb": "Grok 4.6 is a model from SpaceXAI with frontier performance on coding, knowledge work, and STEM. It is succeeded by [Grok 4.7](/x-ai/grok-4.7).",
      "contextWindow": 500000,
      "maxOutput": 450000,
      "cost": {
        "input": 4,
        "output": 12,
        "cacheRead": 1,
        "webSearch": 0.01,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 8,
            "output": 24,
            "cacheRead": 2
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-08-12",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 157.79,
        "accessibility": "API access",
        "gpqa": 0.94,
        "gpqaStderr": 0.0145,
        "aime": 0.992,
        "aimeStderr": 0.0047,
        "simpleQA": 0.493,
        "simpleQAStderr": 0.0158
      }
    },
    {
      "id": "x-ai/grok-4.6:batch",
      "orId": "x-ai/grok-4.6:batch",
      "vendor": "x-ai",
      "model": "grok-4.6:batch",
      "name": "SpaceXAI: Grok 4.6 (batch)",
      "blurb": "Grok 4.6 is a model from SpaceXAI with frontier performance on coding, knowledge work, and STEM. It is succeeded by [Grok 4.7](/x-ai/grok-4.7).",
      "contextWindow": 500000,
      "maxOutput": 450000,
      "cost": {
        "input": 3.2,
        "output": 9.6,
        "cacheRead": 0.8,
        "webSearch": 0.01,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 6.4,
            "output": 19.2,
            "cacheRead": 1.6
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-08-12"
    },
    {
      "id": "x-ai/grok-4.7",
      "orId": "x-ai/grok-4.7",
      "vendor": "x-ai",
      "model": "grok-4.7",
      "name": "SpaceXAI: Grok 4.7",
      "blurb": "Grok 4.7 is SpaceXAI's flagship model for coding, agentic tasks, and knowledge work, succeeding Grok 4.6. It is particularly strong at long-running software engineering tasks, verifying its own work, and...",
      "contextWindow": 500000,
      "maxOutput": 450000,
      "cost": {
        "input": 4,
        "output": 12,
        "cacheRead": 1,
        "webSearch": 0.01,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 8,
            "output": 24,
            "cacheRead": 2
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-21",
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 156.18,
        "accessibility": "API access",
        "gpqa": 0.927,
        "gpqaStderr": 0.0155,
        "aime": 0.981,
        "aimeStderr": 0.0143,
        "simpleQA": 0.56,
        "simpleQAStderr": 0.0157
      }
    },
    {
      "id": "x-ai/grok-4.7:batch",
      "orId": "x-ai/grok-4.7:batch",
      "vendor": "x-ai",
      "model": "grok-4.7:batch",
      "name": "SpaceXAI: Grok 4.7 (batch)",
      "blurb": "Grok 4.7 is SpaceXAI's flagship model for coding, agentic tasks, and knowledge work, succeeding Grok 4.6. It is particularly strong at long-running software engineering tasks, verifying its own work, and...",
      "contextWindow": 500000,
      "maxOutput": 450000,
      "cost": {
        "input": 3.2,
        "output": 9.6,
        "cacheRead": 0.8,
        "webSearch": 0.01,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 6.4,
            "output": 19.2,
            "cacheRead": 1.6
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "xhigh",
          "high",
          "medium",
          "low"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-09-21"
    },
    {
      "id": "x-ai/grok-build-0.1",
      "orId": "x-ai/grok-build-0.1",
      "vendor": "x-ai",
      "model": "grok-build-0.1",
      "name": "SpaceXAI: Grok Build 0.1",
      "blurb": "Grok Build 0.1 is SpaceXAI’s fast coding model trained specifically for agentic software engineering workflows. It supports text and image inputs with text output, and is optimized for interactive coding...",
      "contextWindow": 256000,
      "maxOutput": 230400,
      "cost": {
        "input": 2,
        "output": 4,
        "cacheRead": 0.4,
        "webSearch": 0.01,
        "tiers": [
          {
            "minPromptTokens": 200000,
            "input": 4,
            "output": 8,
            "cacheRead": 0.8
          }
        ]
      },
      "input": [
        "text",
        "image",
        "file"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true
      },
      "tools": true,
      "released": "2026-05-20"
    },
    {
      "id": "xiaomi/mimo-v2.5",
      "orId": "xiaomi/mimo-v2.5",
      "vendor": "xiaomi",
      "model": "mimo-v2.5",
      "name": "Xiaomi: MiMo-V2.5",
      "blurb": "MiMo-V2.5 is a native omnimodal model by Xiaomi. It delivers Pro-level agentic performance at roughly half the inference cost, while surpassing MiMo-V2-Omni in multimodal perception across image and video understanding...",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "cost": {
        "input": 0.8,
        "output": 4,
        "cacheRead": 0.16
      },
      "input": [
        "text",
        "audio",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-22",
      "openWeights": {
        "huggingFace": "XiaomiMiMo/MiMo-V2.5"
      }
    },
    {
      "id": "xiaomi/mimo-v2.5-pro",
      "orId": "xiaomi/mimo-v2.5-pro",
      "vendor": "xiaomi",
      "model": "mimo-v2.5-pro",
      "name": "Xiaomi: MiMo-V2.5-Pro",
      "blurb": "MiMo-V2.5-Pro is Xiaomi’s flagship model, delivering strong performance in general agentic capabilities, complex software engineering, and long-horizon tasks, with top rankings on benchmarks such as ClawEval, GDPVal, and SWE-bench Pro....",
      "contextWindow": 1024000,
      "maxOutput": 131072,
      "cost": {
        "input": 0.87,
        "output": 1.74,
        "cacheRead": 0.0072
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-22",
      "openWeights": {
        "huggingFace": "XiaomiMiMo/MiMo-V2.5-Pro"
      }
    },
    {
      "id": "xiaomi/mimo-v2.6-flash",
      "orId": "xiaomi/mimo-v2.6-flash",
      "vendor": "xiaomi",
      "model": "mimo-v2.6-flash",
      "name": "Xiaomi: MiMo-V2.6-Flash",
      "blurb": "MiMo-V2.6-Flash is an open-source foundation model developed by Xiaomi. Built on a Mixture-of-Experts architecture with 309B total parameters and 15B activated per token, it employs a hybrid attention mechanism for...",
      "contextWindow": 1050000,
      "maxOutput": 131072,
      "cost": {
        "input": 0.28,
        "output": 0.56,
        "cacheRead": 0.0056
      },
      "input": [
        "text",
        "image",
        "video",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-09-21",
      "openWeights": {
        "huggingFace": "XiaomiMiMo/MiMo-V2.6-Flash-RL"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "xiaomi/mimo-v2.6-pro",
      "orId": "xiaomi/mimo-v2.6-pro",
      "vendor": "xiaomi",
      "model": "mimo-v2.6-pro",
      "name": "Xiaomi: MiMo-V2.6-Pro",
      "blurb": "MiMo-V2.6-Pro is the flagship foundation model developed by Xiaomi. Built at a scale of over 1T parameters, it is designed to push the ceiling of capability for the most demanding...",
      "contextWindow": 1050000,
      "maxOutput": 131072,
      "cost": {
        "input": 0.87,
        "output": 1.74,
        "cacheRead": 0.0072
      },
      "input": [
        "text",
        "image",
        "video",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-09-21",
      "openWeights": {
        "huggingFace": "XiaomiMiMo/MiMo-V2.6-Pro-RL"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "xiaomi/mimo-v2.6-pro-ultraspeed",
      "orId": "xiaomi/mimo-v2.6-pro-ultraspeed",
      "vendor": "xiaomi",
      "model": "mimo-v2.6-pro-ultraspeed",
      "name": "Xiaomi: MiMo-V2.6-Pro-UltraSpeed",
      "blurb": "MiMo-V2.6-Pro-UltraSpeed is the fast speed edition of Xiaomi's flagship foundation model, MiMo-V2.6-Pro. Built from the same 1T MiMo-V2.6-Pro checkpoint, it matches the original model in quality while delivering roughly 10x...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 8.7,
        "output": 17.4,
        "cacheRead": 0.072
      },
      "input": [
        "text",
        "image",
        "video",
        "audio"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-09-21"
    },
    {
      "id": "z-ai/glm-4.5",
      "orId": "z-ai/glm-4.5",
      "vendor": "z-ai",
      "model": "glm-4.5",
      "name": "Z.ai: GLM 4.5",
      "blurb": "GLM-4.5 is our latest flagship foundation model, purpose-built for agent-based applications. It leverages a Mixture-of-Experts (MoE) architecture and supports a context length of up to 128k tokens. GLM-4.5 delivers significantly...",
      "contextWindow": 131072,
      "maxOutput": 98304,
      "cost": {
        "input": 1.2,
        "output": 4.4,
        "cacheRead": 0.22
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2024-12-31",
      "released": "2025-07-25",
      "retiresAt": "2026-12-31",
      "openWeights": {
        "huggingFace": "zai-org/GLM-4.5"
      }
    },
    {
      "id": "z-ai/glm-4.5-air",
      "orId": "z-ai/glm-4.5-air",
      "vendor": "z-ai",
      "model": "glm-4.5-air",
      "name": "Z.ai: GLM 4.5 Air",
      "blurb": "GLM-4.5-Air is the lightweight variant of our latest flagship model family, also purpose-built for agent-centric applications. Like GLM-4.5, it adopts the Mixture-of-Experts (MoE) architecture but with a more compact parameter...",
      "contextWindow": 131072,
      "maxOutput": 98304,
      "cost": {
        "input": 0.26,
        "output": 1.7,
        "cacheRead": 0.05
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2024-12-31",
      "released": "2025-07-25",
      "openWeights": {
        "huggingFace": "zai-org/GLM-4.5-Air"
      }
    },
    {
      "id": "z-ai/glm-4.5v",
      "orId": "z-ai/glm-4.5v",
      "vendor": "z-ai",
      "model": "glm-4.5v",
      "name": "Z.ai: GLM 4.5V",
      "blurb": "GLM-4.5V is a vision-language foundation model for multimodal agent applications. Built on a Mixture-of-Experts (MoE) architecture with 106B parameters and 12B activated parameters, it achieves state-of-the-art results in video understanding,...",
      "contextWindow": 65536,
      "maxOutput": 16384,
      "cost": {
        "input": 1.2,
        "output": 3.6,
        "cacheRead": 0.22
      },
      "input": [
        "text",
        "image"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2024-12-31",
      "released": "2025-08-11",
      "retiresAt": "2026-12-31",
      "openWeights": {
        "huggingFace": "zai-org/GLM-4.5V"
      }
    },
    {
      "id": "z-ai/glm-4.6",
      "orId": "z-ai/glm-4.6",
      "vendor": "z-ai",
      "model": "glm-4.6",
      "name": "Z.ai: GLM 4.6",
      "blurb": "Compared with GLM-4.5, this generation brings several key improvements: Longer context window: The context window has been expanded from 128K to 200K tokens, enabling the model to handle more complex...",
      "contextWindow": 204800,
      "maxOutput": 131072,
      "cost": {
        "input": 1,
        "output": 4,
        "cacheRead": 0.2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "knowledgeCutoff": "2025-03-31",
      "released": "2025-09-30",
      "openWeights": {
        "huggingFace": "zai-org/GLM-4.6"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "z-ai/glm-4.6v",
      "orId": "z-ai/glm-4.6v",
      "vendor": "z-ai",
      "model": "glm-4.6v",
      "name": "Z.ai: GLM 4.6V",
      "blurb": "GLM-4.6V is a large multimodal model designed for high-fidelity visual understanding and long-context reasoning across images, documents, and mixed media. It supports up to 128K tokens, processes complex page layouts...",
      "contextWindow": 131072,
      "maxOutput": 32768,
      "cost": {
        "input": 0.6,
        "output": 1.8,
        "cacheRead": 0.11
      },
      "input": [
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2025-12-08",
      "openWeights": {
        "huggingFace": "zai-org/GLM-4.6V"
      }
    },
    {
      "id": "z-ai/glm-4.7",
      "orId": "z-ai/glm-4.7",
      "vendor": "z-ai",
      "model": "glm-4.7",
      "name": "Z.ai: GLM 4.7",
      "blurb": "GLM-4.7 is Z.ai’s latest flagship model, featuring upgrades in two key areas: enhanced programming capabilities and more stable multi-step reasoning/execution. It demonstrates significant improvements in executing complex agent tasks while...",
      "contextWindow": 204800,
      "maxOutput": 131072,
      "cost": {
        "input": 0.8,
        "output": 3.5,
        "cacheRead": 0.16
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2025-12-22",
      "openWeights": {
        "huggingFace": "zai-org/GLM-4.7"
      },
      "regions": [
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 143.28,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.833,
        "gpqaStderr": 0.0237,
        "aime": 0.833,
        "aimeStderr": 0.0477,
        "simpleQA": 0.322,
        "simpleQAStderr": 0.0148
      }
    },
    {
      "id": "z-ai/glm-4.7-flash",
      "orId": "z-ai/glm-4.7-flash",
      "vendor": "z-ai",
      "model": "glm-4.7-flash",
      "name": "Z.ai: GLM 4.7 Flash",
      "blurb": "As a 30B-class SOTA model, GLM-4.7-Flash offers a new option that balances performance and efficiency. It is further optimized for agentic coding use cases, strengthening coding capabilities, long-horizon task planning,...",
      "contextWindow": 200000,
      "maxOutput": 117964,
      "cost": {
        "input": 0.121,
        "output": 0.8
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-01-19",
      "openWeights": {
        "huggingFace": "zai-org/GLM-4.7-Flash"
      },
      "bench": {
        "source": "epoch",
        "gpqa": 0.605,
        "gpqaStderr": 0.0293,
        "aime": 0.583,
        "aimeStderr": 0.0656
      }
    },
    {
      "id": "z-ai/glm-5",
      "orId": "z-ai/glm-5",
      "vendor": "z-ai",
      "model": "glm-5",
      "name": "Z.ai: GLM 5",
      "blurb": "GLM-5 is Z.ai’s flagship open-source foundation model engineered for complex systems design and long-horizon agent workflows. Built for expert developers, it delivers production-grade performance on large-scale programming tasks, rivaling leading...",
      "contextWindow": 204800,
      "maxOutput": 131072,
      "cost": {
        "input": 2,
        "output": 6.4,
        "cacheRead": 0.4
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-02-11",
      "openWeights": {
        "huggingFace": "zai-org/GLM-5"
      },
      "bench": {
        "source": "epoch",
        "eci": 146.48,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.878,
        "gpqaStderr": 0.0234,
        "sweBenchVerified": 0.721,
        "sweBenchVerifiedStderr": 0.0209,
        "frontierMath": 0.164,
        "frontierMathStderr": 0.022,
        "aime": 0.8,
        "aimeStderr": 0.06
      }
    },
    {
      "id": "z-ai/glm-5-turbo",
      "orId": "z-ai/glm-5-turbo",
      "vendor": "z-ai",
      "model": "glm-5-turbo",
      "name": "Z.ai: GLM 5 Turbo",
      "blurb": "GLM-5 Turbo is a new model from Z.ai designed for fast inference and strong performance in agent-driven environments such as OpenClaw scenarios. It is deeply optimized for real-world agent workflows...",
      "contextWindow": 202752,
      "maxOutput": 131072,
      "cost": {
        "input": 2.4,
        "output": 8,
        "cacheRead": 0.48
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-03-15"
    },
    {
      "id": "z-ai/glm-5.1",
      "orId": "z-ai/glm-5.1",
      "vendor": "z-ai",
      "model": "glm-5.1",
      "name": "Z.ai: GLM 5.1",
      "blurb": "GLM-5.1 delivers a major leap in coding capability, with particularly significant gains in handling long-horizon tasks. Unlike previous models built around minute-level interactions, GLM-5.1 can work independently and continuously on...",
      "contextWindow": 204800,
      "maxOutput": 182476,
      "cost": {
        "input": 2.8,
        "output": 8.8,
        "cacheRead": 0.52
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-07",
      "retiresAt": "2026-11-20",
      "openWeights": {
        "huggingFace": "zai-org/GLM-5.1"
      },
      "bench": {
        "source": "epoch",
        "eci": 150.73,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.899,
        "gpqaStderr": 0.0215,
        "sweBenchVerified": 0.742,
        "sweBenchVerifiedStderr": 0.0199,
        "frontierMath": 0.334,
        "frontierMathStderr": 0.0278,
        "aime": 0.933,
        "aimeStderr": 0.0376,
        "simpleQA": 0.34,
        "simpleQAStderr": 0.015
      }
    },
    {
      "id": "z-ai/glm-5.2",
      "orId": "z-ai/glm-5.2",
      "vendor": "z-ai",
      "model": "glm-5.2",
      "name": "Z.ai: GLM 5.2",
      "blurb": "GLM 5.2 is a large-scale reasoning model from Z.ai. It supports text input and output with a 1M-token context window, and is suited for long-horizon agent workflows, project-level software engineering,...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.014,
        "output": 13.6,
        "cacheRead": 0.012
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false,
        "efforts": [
          "xhigh",
          "high",
          "none"
        ],
        "defaultEffort": "high"
      },
      "tools": true,
      "released": "2026-06-16",
      "openWeights": {
        "huggingFace": "zai-org/GLM-5.2"
      },
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 153.03,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.919,
        "gpqaStderr": 0.0161,
        "sweBenchVerified": 0.787,
        "sweBenchVerifiedStderr": 0.0187,
        "aime": 0.864,
        "aimeStderr": 0.0422,
        "simpleQA": 0.342,
        "simpleQAStderr": 0.015
      }
    },
    {
      "id": "z-ai/glm-5.3",
      "orId": "z-ai/glm-5.3",
      "vendor": "z-ai",
      "model": "glm-5.3",
      "name": "Z.ai: GLM 5.3",
      "blurb": "GLM-5.3 is a large-scale reasoning model from Z.ai, built for complex software engineering and long-horizon agent tasks. It supports text input and output with a 1M-token context window, and improves...",
      "contextWindow": 1048576,
      "maxOutput": 943718,
      "cost": {
        "input": 0.1398,
        "output": 2.8,
        "cacheRead": 0.1378
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-08-18",
      "openWeights": {
        "huggingFace": "zai-org/GLM-5.3"
      },
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 157.1,
        "accessibility": "Open weights (restricted use)",
        "gpqa": 0.909,
        "gpqaStderr": 0.0159,
        "aime": 0.911,
        "aimeStderr": 0.0345,
        "simpleQA": 0.41,
        "simpleQAStderr": 0.0156
      }
    },
    {
      "id": "z-ai/glm-5.3-flash",
      "orId": "z-ai/glm-5.3-flash",
      "vendor": "z-ai",
      "model": "glm-5.3-flash",
      "name": "Z.ai: GLM 5.3 Flash",
      "blurb": "GLM-5.3-Flash is a native multimodal model from Z.ai. It is suited for efficient coding and long-horizon agent tasks. Its hybrid sparse and linear attention architecture maintains accurate long-context behavior while...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0.0798,
        "output": 0.1568,
        "cacheRead": 0.06
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-08-26",
      "openWeights": {
        "huggingFace": "zai-org/GLM-5.3-Flash"
      },
      "regions": [
        "eu",
        "us"
      ],
      "bench": {
        "source": "epoch",
        "eci": 153.82,
        "accessibility": "Open weights (unrestricted)",
        "gpqa": 0.902,
        "gpqaStderr": 0.0172,
        "aime": 0.939,
        "aimeStderr": 0.0216
      }
    },
    {
      "id": "z-ai/glm-5.3-flash:batch",
      "orId": "z-ai/glm-5.3-flash:batch",
      "vendor": "z-ai",
      "model": "glm-5.3-flash:batch",
      "name": "Z.ai: GLM 5.3 Flash (batch)",
      "blurb": "GLM-5.3-Flash is a native multimodal model from Z.ai. It is suited for efficient coding and long-horizon agent tasks. Its hybrid sparse and linear attention architecture maintains accurate long-context behavior while...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0.1,
        "output": 0.36,
        "cacheRead": 0.02
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-08-26",
      "openWeights": {
        "huggingFace": "zai-org/GLM-5.3-Flash"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "z-ai/glm-5.3-flashx",
      "orId": "z-ai/glm-5.3-flashx",
      "vendor": "z-ai",
      "model": "glm-5.3-flashx",
      "name": "Z.ai: GLM 5.3 FlashX",
      "blurb": "GLM-5.3-FlashX is the high-speed variant of Z.ai's GLM-5.3-Flash, a native multimodal model delivering inference speeds of up to 200 tokens/s. Built on the same hybrid sparse and linear attention architecture...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0.74,
        "output": 2.5,
        "cacheRead": 0.15
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-09-18"
    },
    {
      "id": "z-ai/glm-5.3-flashx-legal",
      "orId": "z-ai/glm-5.3-flashx-legal",
      "vendor": "z-ai",
      "model": "glm-5.3-flashx-legal",
      "name": "Z.ai: GLM 5.3 FlashX Legal",
      "blurb": "GLM-5.3-FlashX-Legal (GLM-Legal) is Z.ai's legal-domain variant of GLM-5.3-FlashX, a native multimodal model that accepts text, image, and video input and produces text. It adds legal-domain mid-training and post-training with reinforcement...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0.74,
        "output": 2.5,
        "cacheRead": 0.15
      },
      "input": [
        "text",
        "image",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-11-02"
    },
    {
      "id": "z-ai/glm-5.3-prime",
      "orId": "z-ai/glm-5.3-prime",
      "vendor": "z-ai",
      "model": "glm-5.3-prime",
      "name": "Z.ai: GLM 5.3 Prime",
      "blurb": "GLM-5.3-Prime is the high-speed variant of Z.ai's GLM-5.3, inheriting its full capabilities while delivering 1.5–2× the output throughput through inference acceleration. It supports text input and output with a 1M-token...",
      "contextWindow": 1000000,
      "maxOutput": 131072,
      "cost": {
        "input": 5.6,
        "output": 17.6,
        "cacheRead": 1.12
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-09-23"
    },
    {
      "id": "z-ai/glm-5.3:batch",
      "orId": "z-ai/glm-5.3:batch",
      "vendor": "z-ai",
      "model": "glm-5.3:batch",
      "name": "Z.ai: GLM 5.3 (batch)",
      "blurb": "GLM-5.3 is a large-scale reasoning model from Z.ai, built for complex software engineering and long-horizon agent tasks. It supports text input and output with a 1M-token context window, and improves...",
      "contextWindow": 1048576,
      "maxOutput": 131072,
      "cost": {
        "input": 0.9,
        "output": 4,
        "cacheRead": 0.2
      },
      "input": [
        "text"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": true,
        "efforts": [
          "max",
          "high",
          "low"
        ],
        "defaultEffort": "max"
      },
      "tools": true,
      "released": "2026-08-18",
      "openWeights": {
        "huggingFace": "zai-org/GLM-5.3"
      },
      "regions": [
        "us"
      ]
    },
    {
      "id": "z-ai/glm-5v-turbo",
      "orId": "z-ai/glm-5v-turbo",
      "vendor": "z-ai",
      "model": "glm-5v-turbo",
      "name": "Z.ai: GLM 5V Turbo",
      "blurb": "GLM-5V-Turbo is Z.ai’s first native multimodal agent foundation model, built for vision-based coding and agent-driven tasks. It natively handles image, video, and text inputs, excels at long-horizon planning, complex coding,...",
      "contextWindow": 202752,
      "maxOutput": 131072,
      "cost": {
        "input": 2.4,
        "output": 8,
        "cacheRead": 0.48
      },
      "input": [
        "image",
        "text",
        "video"
      ],
      "output": [
        "text"
      ],
      "reasoning": {
        "mandatory": false
      },
      "tools": true,
      "released": "2026-04-01"
    }
  ],
  "llmSource": {
    "text": "Prices are 00 AI-token prices — final, per million tokens. Bring your own provider key to pay provider rates directly.",
    "url": "https://0-0.chat",
    "benchmarks": {
      "text": "Benchmark scores by Epoch AI, CC BY 4.0",
      "url": "https://epoch.ai/data/ai-benchmarking-dashboard",
      "license": "CC-BY-4.0",
      "files": [
        "epoch_capabilities_index",
        "gpqa_diamond",
        "swe_bench_verified",
        "frontiermath",
        "math_level_5",
        "otis_mock_aime_2024_2025",
        "simpleqa_verified"
      ],
      "fetchedAt": "2026-11-16T06:40:06.761Z"
    }
  },
  "localLlms": [
    {
      "id": "llama3.1",
      "label": "llama3.1",
      "description": "Llama 3.1 is a new state-of-the-art model from Meta available in 8B, 70B and 405B parameter sizes.",
      "sizes": [
        "8b",
        "70b",
        "405b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/llama3.1",
      "install": "ollama pull llama3.1"
    },
    {
      "id": "nomic-embed-text",
      "label": "nomic-embed-text",
      "description": "A high-performing open embedding model with a large token context window.",
      "sizes": [],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/nomic-embed-text",
      "install": "ollama pull nomic-embed-text"
    },
    {
      "id": "deepseek-r1",
      "label": "deepseek-r1",
      "description": "DeepSeek-R1 is a family of open reasoning models with performance approaching that of leading models, such as O3 and Gemini 2.5 Pro.",
      "sizes": [
        "1.5b",
        "7b",
        "8b",
        "14b",
        "32b",
        "70b",
        "671b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/deepseek-r1",
      "install": "ollama pull deepseek-r1"
    },
    {
      "id": "llama3.2",
      "label": "llama3.2",
      "description": "Meta's Llama 3.2 goes small with 1B and 3B models.",
      "sizes": [
        "1b",
        "3b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/llama3.2",
      "install": "ollama pull llama3.2"
    },
    {
      "id": "qwen2.5",
      "label": "qwen2.5",
      "description": "Qwen2.5 models are pretrained on Alibaba's latest large-scale dataset, encompassing up to 18 trillion tokens. The model supports up to 128K tokens and has multilingual support.",
      "sizes": [
        "0.5b",
        "1.5b",
        "3b",
        "7b",
        "14b",
        "32b",
        "72b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/qwen2.5",
      "install": "ollama pull qwen2.5"
    },
    {
      "id": "qwen3",
      "label": "qwen3",
      "description": "Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.",
      "sizes": [
        "0.6b",
        "1.7b",
        "4b",
        "8b",
        "14b",
        "30b",
        "32b",
        "235b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/qwen3",
      "install": "ollama pull qwen3"
    },
    {
      "id": "gemma3",
      "label": "gemma3",
      "description": "The current, most capable model that runs on a single GPU.",
      "sizes": [
        "270m",
        "1b",
        "4b",
        "12b",
        "27b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/gemma3",
      "install": "ollama pull gemma3"
    },
    {
      "id": "gemma2",
      "label": "gemma2",
      "description": "Google Gemma 2 is a high-performing and efficient model available in three sizes: 2B, 9B, and 27B.",
      "sizes": [
        "2b",
        "9b",
        "27b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/gemma2",
      "install": "ollama pull gemma2"
    },
    {
      "id": "mistral",
      "label": "mistral",
      "description": "The 7B model released by Mistral AI, updated to version 0.3.",
      "sizes": [
        "7b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/mistral",
      "install": "ollama pull mistral"
    },
    {
      "id": "gemma4",
      "label": "gemma4",
      "description": "Gemma 4 models are designed to deliver frontier-level performance at each size. They are well-suited for reasoning, agentic workflows, coding, and multimodal understanding.",
      "sizes": [
        "12b",
        "26b",
        "31b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "audio",
        "cloud",
        "e2b",
        "e4b"
      ],
      "url": "https://ollama.com/library/gemma4",
      "install": "ollama pull gemma4"
    },
    {
      "id": "llama3",
      "label": "llama3",
      "description": "Meta Llama 3: The most capable openly available LLM to date",
      "sizes": [
        "8b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/llama3",
      "install": "ollama pull llama3"
    },
    {
      "id": "qwen3.5",
      "label": "qwen3.5",
      "description": "Qwen 3.5 is a family of open-source multimodal models that delivers exceptional utility and performance.",
      "sizes": [
        "0.8b",
        "2b",
        "4b",
        "9b",
        "27b",
        "35b",
        "122b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/qwen3.5",
      "install": "ollama pull qwen3.5"
    },
    {
      "id": "qwen2.5-coder",
      "label": "qwen2.5-coder",
      "description": "The latest series of Code-Specific Qwen models, with significant improvements in code generation, code reasoning, and code fixing.",
      "sizes": [
        "0.5b",
        "1.5b",
        "3b",
        "7b",
        "14b",
        "32b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/qwen2.5-coder",
      "install": "ollama pull qwen2.5-coder"
    },
    {
      "id": "phi3",
      "label": "phi3",
      "description": "Phi-3 is a family of lightweight 3B (Mini) and 14B (Medium) state-of-the-art open models by Microsoft.",
      "sizes": [
        "3.8b",
        "14b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/phi3",
      "install": "ollama pull phi3"
    },
    {
      "id": "mxbai-embed-large",
      "label": "mxbai-embed-large",
      "description": "State-of-the-art large embedding model from mixedbread.ai",
      "sizes": [
        "335m"
      ],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/mxbai-embed-large",
      "install": "ollama pull mxbai-embed-large"
    },
    {
      "id": "llava",
      "label": "llava",
      "description": "🌋 LLaVA is a novel end-to-end trained large multimodal model that combines a vision encoder and Vicuna for general-purpose visual and language understanding. Updated to version 1.6.",
      "sizes": [
        "7b",
        "13b",
        "34b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/llava",
      "install": "ollama pull llava"
    },
    {
      "id": "gpt-oss",
      "label": "gpt-oss",
      "description": "OpenAI’s open-weight models designed for powerful reasoning, agentic tasks, and versatile developer use cases.",
      "sizes": [
        "20b",
        "120b"
      ],
      "capabilities": [
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/gpt-oss",
      "install": "ollama pull gpt-oss"
    },
    {
      "id": "qwen3-coder",
      "label": "qwen3-coder",
      "description": "Alibaba's performant long context models for agentic and coding tasks.",
      "sizes": [
        "30b",
        "480b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/qwen3-coder",
      "install": "ollama pull qwen3-coder"
    },
    {
      "id": "glm-ocr",
      "label": "glm-ocr",
      "description": "GLM-OCR is a multimodal OCR model for complex document understanding, built on the GLM-V encoder–decoder architecture.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools"
      ],
      "url": "https://ollama.com/library/glm-ocr",
      "install": "ollama pull glm-ocr"
    },
    {
      "id": "gemma",
      "label": "gemma",
      "description": "Gemma is a family of lightweight, state-of-the-art open models built by Google DeepMind. Updated to version 1.1",
      "sizes": [
        "2b",
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/gemma",
      "install": "ollama pull gemma"
    },
    {
      "id": "bge-m3",
      "label": "bge-m3",
      "description": "BGE-M3 is a new model from BAAI distinguished for its versatility in Multi-Functionality, Multi-Linguality, and Multi-Granularity.",
      "sizes": [
        "567m"
      ],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/bge-m3",
      "install": "ollama pull bge-m3"
    },
    {
      "id": "qwen",
      "label": "qwen",
      "description": "Qwen 1.5 is a series of large language models by Alibaba Cloud spanning from 0.5B to 110B parameters",
      "sizes": [
        "0.5b",
        "1.8b",
        "4b",
        "7b",
        "14b",
        "32b",
        "72b",
        "110b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/qwen",
      "install": "ollama pull qwen"
    },
    {
      "id": "codellama",
      "label": "codellama",
      "description": "A large language model that can use text prompts to generate and discuss code.",
      "sizes": [
        "7b",
        "13b",
        "34b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/codellama",
      "install": "ollama pull codellama"
    },
    {
      "id": "phi4",
      "label": "phi4",
      "description": "Phi-4 is a 14B parameter, state-of-the-art open model from Microsoft.",
      "sizes": [
        "14b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/phi4",
      "install": "ollama pull phi4"
    },
    {
      "id": "llama2",
      "label": "llama2",
      "description": "Llama 2 is a collection of foundation language models ranging from 7B to 70B parameters.",
      "sizes": [
        "7b",
        "13b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/llama2",
      "install": "ollama pull llama2"
    },
    {
      "id": "qwen3.6",
      "label": "qwen3.6",
      "description": "Qwen3.6 delivers substantial upgrades in agentic coding and thinking preservation than previous Qwen models.",
      "sizes": [
        "27b",
        "35b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/qwen3.6",
      "install": "ollama pull qwen3.6"
    },
    {
      "id": "qwen3-vl",
      "label": "qwen3-vl",
      "description": "The most powerful vision-language model in the Qwen model family to date.",
      "sizes": [
        "2b",
        "4b",
        "8b",
        "30b",
        "32b",
        "235b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/qwen3-vl",
      "install": "ollama pull qwen3-vl"
    },
    {
      "id": "tinyllama",
      "label": "tinyllama",
      "description": "The TinyLlama project is an open endeavor to train a compact 1.1B Llama model on 3 trillion tokens.",
      "sizes": [
        "1.1b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/tinyllama",
      "install": "ollama pull tinyllama"
    },
    {
      "id": "qwen3.8",
      "label": "qwen3.8",
      "description": "Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks.",
      "sizes": [
        "27b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/qwen3.8",
      "install": "ollama pull qwen3.8"
    },
    {
      "id": "qwen2",
      "label": "qwen2",
      "description": "Qwen2 is a new series of large language models from Alibaba group",
      "sizes": [
        "0.5b",
        "1.5b",
        "7b",
        "72b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/qwen2",
      "install": "ollama pull qwen2"
    },
    {
      "id": "qwen2.5vl",
      "label": "qwen2.5vl",
      "description": "Flagship vision-language model of Qwen and also a significant leap from the previous Qwen2-VL.",
      "sizes": [
        "3b",
        "7b",
        "32b",
        "72b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/qwen2.5vl",
      "install": "ollama pull qwen2.5vl"
    },
    {
      "id": "mistral-nemo",
      "label": "mistral-nemo",
      "description": "A state-of-the-art 12B model with 128k context length, built by Mistral AI in collaboration with NVIDIA.",
      "sizes": [
        "12b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/mistral-nemo",
      "install": "ollama pull mistral-nemo"
    },
    {
      "id": "minicpm-v",
      "label": "minicpm-v",
      "description": "A series of multimodal LLMs (MLLMs) designed for vision-language understanding.",
      "sizes": [
        "8b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/minicpm-v",
      "install": "ollama pull minicpm-v"
    },
    {
      "id": "llama3.2-vision",
      "label": "llama3.2-vision",
      "description": "Llama 3.2 Vision is a collection of instruction-tuned image reasoning generative models in 11B and 90B sizes.",
      "sizes": [
        "11b",
        "90b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/llama3.2-vision",
      "install": "ollama pull llama3.2-vision"
    },
    {
      "id": "deepseek-coder",
      "label": "deepseek-coder",
      "description": "DeepSeek Coder is a capable coding model trained on two trillion code and natural language tokens.",
      "sizes": [
        "1.3b",
        "6.7b",
        "33b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/deepseek-coder",
      "install": "ollama pull deepseek-coder"
    },
    {
      "id": "qwen3-embedding",
      "label": "qwen3-embedding",
      "description": "Building upon the foundational models of the Qwen3 series, Qwen3 Embedding provides a comprehensive range of text embeddings models in various sizes",
      "sizes": [
        "0.6b",
        "4b",
        "8b"
      ],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/qwen3-embedding",
      "install": "ollama pull qwen3-embedding"
    },
    {
      "id": "llama3.3",
      "label": "llama3.3",
      "description": "New state of the art 70B model. Llama 3.3 70B offers similar performance compared to the Llama 3.1 405B model.",
      "sizes": [
        "70b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/llama3.3",
      "install": "ollama pull llama3.3"
    },
    {
      "id": "dolphin3",
      "label": "dolphin3",
      "description": "Dolphin 3.0 Llama 3.1 8B 🐬 is the next generation of the Dolphin series of instruct-tuned models designed to be the ultimate general purpose local model, enabling coding, math, agentic, function calling, and general use cases.",
      "sizes": [
        "8b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/dolphin3",
      "install": "ollama pull dolphin3"
    },
    {
      "id": "smollm2",
      "label": "smollm2",
      "description": "SmolLM2 is a family of compact language models available in three size: 135M, 360M, and 1.7B parameters.",
      "sizes": [
        "135m",
        "360m",
        "1.7b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/smollm2",
      "install": "ollama pull smollm2"
    },
    {
      "id": "all-minilm",
      "label": "all-minilm",
      "description": "Embedding models on very large sentence level datasets.",
      "sizes": [
        "22m",
        "33m"
      ],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/all-minilm",
      "install": "ollama pull all-minilm"
    },
    {
      "id": "deepseek-v3",
      "label": "deepseek-v3",
      "description": "A strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token.",
      "sizes": [
        "671b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/deepseek-v3",
      "install": "ollama pull deepseek-v3"
    },
    {
      "id": "olmo2",
      "label": "olmo2",
      "description": "OLMo 2 is a new family of 7B and 13B models trained on up to 5T tokens. These models are on par with or better than equivalently sized fully open models, and competitive with open-weight models such as Llama 3.1 on English academic benchmarks.",
      "sizes": [
        "7b",
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/olmo2",
      "install": "ollama pull olmo2"
    },
    {
      "id": "deepseek-coder-v2",
      "label": "deepseek-coder-v2",
      "description": "An open-source Mixture-of-Experts code language model that achieves performance comparable to GPT4-Turbo in code-specific tasks.",
      "sizes": [
        "16b",
        "236b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/deepseek-coder-v2",
      "install": "ollama pull deepseek-coder-v2"
    },
    {
      "id": "codegemma",
      "label": "codegemma",
      "description": "CodeGemma is a collection of powerful, lightweight models that can perform a variety of coding tasks like fill-in-the-middle code completion, code generation, natural language understanding, mathematical reasoning, and instruction following.",
      "sizes": [
        "2b",
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/codegemma",
      "install": "ollama pull codegemma"
    },
    {
      "id": "snowflake-arctic-embed",
      "label": "snowflake-arctic-embed",
      "description": "A suite of text embedding models by Snowflake, optimized for performance.",
      "sizes": [
        "22m",
        "33m",
        "110m",
        "137m",
        "335m"
      ],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/snowflake-arctic-embed",
      "install": "ollama pull snowflake-arctic-embed"
    },
    {
      "id": "mistral-small",
      "label": "mistral-small",
      "description": "Mistral Small 3 sets a new benchmark in the “small” Large Language Models category below 70B.",
      "sizes": [
        "22b",
        "24b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/mistral-small",
      "install": "ollama pull mistral-small"
    },
    {
      "id": "orca-mini",
      "label": "orca-mini",
      "description": "A general-purpose model ranging from 3 billion parameters to 70 billion, suitable for entry-level hardware.",
      "sizes": [
        "3b",
        "7b",
        "13b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/orca-mini",
      "install": "ollama pull orca-mini"
    },
    {
      "id": "granite3.1-moe",
      "label": "granite3.1-moe",
      "description": "The IBM Granite 1B and 3B models are long-context mixture of experts (MoE) Granite models from IBM designed for low latency usage.",
      "sizes": [
        "1b",
        "3b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/granite3.1-moe",
      "install": "ollama pull granite3.1-moe"
    },
    {
      "id": "starcoder2",
      "label": "starcoder2",
      "description": "StarCoder2 is the next generation of transparently trained open code LLMs that comes in three sizes: 3B, 7B and 15B parameters.",
      "sizes": [
        "3b",
        "7b",
        "15b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/starcoder2",
      "install": "ollama pull starcoder2"
    },
    {
      "id": "nemotron-3-super",
      "label": "nemotron-3-super",
      "description": "NVIDIA Nemotron 3 Super is a 120B open MoE model activating just 12B parameters to deliver maximum compute efficiency and accuracy for complex multi-agent applications.",
      "sizes": [
        "120b"
      ],
      "capabilities": [
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/nemotron-3-super",
      "install": "ollama pull nemotron-3-super"
    },
    {
      "id": "mixtral",
      "label": "mixtral",
      "description": "A set of Mixture of Experts (MoE) model with open weights by Mistral AI in 8x7b and 8x22b parameter sizes.",
      "sizes": [],
      "capabilities": [
        "tools",
        "8x7b",
        "8x22b"
      ],
      "url": "https://ollama.com/library/mixtral",
      "install": "ollama pull mixtral"
    },
    {
      "id": "translategemma",
      "label": "translategemma",
      "description": "A new collection of open translation models built on Gemma 3, helping people communicate across 55 languages.",
      "sizes": [
        "4b",
        "12b",
        "27b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/translategemma",
      "install": "ollama pull translategemma"
    },
    {
      "id": "embeddinggemma",
      "label": "embeddinggemma",
      "description": "EmbeddingGemma is a 300M parameter embedding model from Google.",
      "sizes": [
        "300m"
      ],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/embeddinggemma",
      "install": "ollama pull embeddinggemma"
    },
    {
      "id": "llama2-uncensored",
      "label": "llama2-uncensored",
      "description": "Uncensored Llama 2 model by George Sung and Jarrad Hope.",
      "sizes": [
        "7b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/llama2-uncensored",
      "install": "ollama pull llama2-uncensored"
    },
    {
      "id": "devstral-small-2",
      "label": "devstral-small-2",
      "description": "24B model that excels at using tools to explore codebases, editing multiple files and power software engineering agents.",
      "sizes": [
        "24b"
      ],
      "capabilities": [
        "vision",
        "tools"
      ],
      "url": "https://ollama.com/library/devstral-small-2",
      "install": "ollama pull devstral-small-2"
    },
    {
      "id": "falcon3",
      "label": "falcon3",
      "description": "A family of efficient AI models under 10B parameters performant in science, math, and coding through innovative training techniques.",
      "sizes": [
        "1b",
        "3b",
        "7b",
        "10b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/falcon3",
      "install": "ollama pull falcon3"
    },
    {
      "id": "mistral-small3.2",
      "label": "mistral-small3.2",
      "description": "An update to Mistral Small that improves on function calling, instruction following, and less repetition errors.",
      "sizes": [
        "24b"
      ],
      "capabilities": [
        "vision",
        "tools"
      ],
      "url": "https://ollama.com/library/mistral-small3.2",
      "install": "ollama pull mistral-small3.2"
    },
    {
      "id": "gemma3n",
      "label": "gemma3n",
      "description": "Gemma 3n models are designed for efficient execution on everyday devices such as laptops, tablets or phones.",
      "sizes": [],
      "capabilities": [
        "vision",
        "e2b",
        "e4b"
      ],
      "url": "https://ollama.com/library/gemma3n",
      "install": "ollama pull gemma3n"
    },
    {
      "id": "smollm",
      "label": "smollm",
      "description": "🪐 A family of small models with 135M, 360M, and 1.7B parameters, trained on a new high-quality dataset.",
      "sizes": [
        "135m",
        "360m",
        "1.7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/smollm",
      "install": "ollama pull smollm"
    },
    {
      "id": "sqlcoder",
      "label": "sqlcoder",
      "description": "SQLCoder is a code completion model fined-tuned on StarCoder for SQL generation tasks",
      "sizes": [
        "7b",
        "15b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/sqlcoder",
      "install": "ollama pull sqlcoder"
    },
    {
      "id": "llava-llama3",
      "label": "llava-llama3",
      "description": "A LLaVA model fine-tuned from Llama 3 Instruct with better scores in several benchmarks.",
      "sizes": [
        "8b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/llava-llama3",
      "install": "ollama pull llava-llama3"
    },
    {
      "id": "qwq",
      "label": "qwq",
      "description": "QwQ is the reasoning model of the Qwen series.",
      "sizes": [
        "32b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/qwq",
      "install": "ollama pull qwq"
    },
    {
      "id": "dolphin-llama3",
      "label": "dolphin-llama3",
      "description": "Dolphin 2.9 is a new model with 8B and 70B sizes by Eric Hartford based on Llama 3 that has a variety of instruction, conversational, and coding skills.",
      "sizes": [
        "8b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/dolphin-llama3",
      "install": "ollama pull dolphin-llama3"
    },
    {
      "id": "qwen3-coder-next",
      "label": "qwen3-coder-next",
      "description": "Qwen3-Coder-Next is a coding-focused language model from Alibaba's Qwen team, optimized for agentic coding workflows and local development.",
      "sizes": [],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/qwen3-coder-next",
      "install": "ollama pull qwen3-coder-next"
    },
    {
      "id": "moondream",
      "label": "moondream",
      "description": "moondream2 is a small vision language model designed to run efficiently on edge devices.",
      "sizes": [
        "1.8b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/moondream",
      "install": "ollama pull moondream"
    },
    {
      "id": "glm-4.7-flash",
      "label": "glm-4.7-flash",
      "description": "As the strongest model in the 30B class, GLM-4.7-Flash offers a new option for lightweight deployment that balances performance and efficiency.",
      "sizes": [],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/glm-4.7-flash",
      "install": "ollama pull glm-4.7-flash"
    },
    {
      "id": "cogito",
      "label": "cogito",
      "description": "Cogito v1 Preview is a family of hybrid reasoning models by Deep Cogito that outperform the best available open models of the same size, including counterparts from LLaMA, DeepSeek, and Qwen across most standard benchmarks.",
      "sizes": [
        "3b",
        "8b",
        "14b",
        "32b",
        "70b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/cogito",
      "install": "ollama pull cogito"
    },
    {
      "id": "starcoder",
      "label": "starcoder",
      "description": "StarCoder is a code generation model trained on 80&#43; programming languages.",
      "sizes": [
        "1b",
        "3b",
        "7b",
        "15b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/starcoder",
      "install": "ollama pull starcoder"
    },
    {
      "id": "dolphin-mixtral",
      "label": "dolphin-mixtral",
      "description": "Uncensored, 8x7b and 8x22b fine-tuned models based on the Mixtral mixture of experts models that excels at coding tasks. Created by Eric Hartford.",
      "sizes": [],
      "capabilities": [
        "8x7b",
        "8x22b"
      ],
      "url": "https://ollama.com/library/dolphin-mixtral",
      "install": "ollama pull dolphin-mixtral"
    },
    {
      "id": "wizardcoder",
      "label": "wizardcoder",
      "description": "State-of-the-art code generation model",
      "sizes": [
        "33b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/wizardcoder",
      "install": "ollama pull wizardcoder"
    },
    {
      "id": "llama4",
      "label": "llama4",
      "description": "Meta's latest collection of multimodal models.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "16x17b",
        "128x17b"
      ],
      "url": "https://ollama.com/library/llama4",
      "install": "ollama pull llama4"
    },
    {
      "id": "dolphin-mistral",
      "label": "dolphin-mistral",
      "description": "The uncensored Dolphin model based on Mistral that excels at coding tasks. Updated to version 2.8.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/dolphin-mistral",
      "install": "ollama pull dolphin-mistral"
    },
    {
      "id": "openhermes",
      "label": "openhermes",
      "description": "OpenHermes 2.5 is a 7B model fine-tuned by Teknium on Mistral with fully open datasets.",
      "sizes": [],
      "capabilities": [],
      "url": "https://ollama.com/library/openhermes",
      "install": "ollama pull openhermes"
    },
    {
      "id": "hermes3",
      "label": "hermes3",
      "description": "Hermes 3 is the latest version of the flagship Hermes series of LLMs by Nous Research",
      "sizes": [
        "3b",
        "8b",
        "70b",
        "405b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/hermes3",
      "install": "ollama pull hermes3"
    },
    {
      "id": "phi4-reasoning",
      "label": "phi4-reasoning",
      "description": "Phi 4 reasoning and reasoning plus are 14-billion parameter open-weight reasoning models that rival much larger models on complex reasoning tasks.",
      "sizes": [
        "14b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/phi4-reasoning",
      "install": "ollama pull phi4-reasoning"
    },
    {
      "id": "phi4-mini",
      "label": "phi4-mini",
      "description": "Phi-4-mini brings significant enhancements in multilingual support, reasoning, and mathematics, and now, the long-awaited function calling feature is finally supported.",
      "sizes": [
        "3.8b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/phi4-mini",
      "install": "ollama pull phi4-mini"
    },
    {
      "id": "granite-code",
      "label": "granite-code",
      "description": "A family of open foundation models by IBM for Code Intelligence",
      "sizes": [
        "3b",
        "8b",
        "20b",
        "34b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/granite-code",
      "install": "ollama pull granite-code"
    },
    {
      "id": "dolphin-phi",
      "label": "dolphin-phi",
      "description": "2.7B uncensored Dolphin model by Eric Hartford, based on the Phi language model by Microsoft Research.",
      "sizes": [
        "2.7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/dolphin-phi",
      "install": "ollama pull dolphin-phi"
    },
    {
      "id": "ministral-3",
      "label": "ministral-3",
      "description": "The Ministral 3 family is designed for edge deployment, capable of running on a wide range of hardware.",
      "sizes": [
        "3b",
        "8b",
        "14b"
      ],
      "capabilities": [
        "vision",
        "tools"
      ],
      "url": "https://ollama.com/library/ministral-3",
      "install": "ollama pull ministral-3"
    },
    {
      "id": "yi",
      "label": "yi",
      "description": "Yi 1.5 is a high-performing, bilingual language model.",
      "sizes": [
        "6b",
        "9b",
        "34b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/yi",
      "install": "ollama pull yi"
    },
    {
      "id": "granite4",
      "label": "granite4",
      "description": "Granite 4 features improved instruction following (IF) and tool-calling capabilities, making them more effective in enterprise applications.",
      "sizes": [
        "350m",
        "1b",
        "3b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/granite4",
      "install": "ollama pull granite4"
    },
    {
      "id": "command-r",
      "label": "command-r",
      "description": "Command R is a Large Language Model optimized for conversational interaction and long context tasks.",
      "sizes": [
        "35b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/command-r",
      "install": "ollama pull command-r"
    },
    {
      "id": "phi",
      "label": "phi",
      "description": "Phi-2: a 2.7B language model by Microsoft Research that demonstrates outstanding reasoning and language understanding capabilities.",
      "sizes": [
        "2.7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/phi",
      "install": "ollama pull phi"
    },
    {
      "id": "magistral",
      "label": "magistral",
      "description": "Magistral is a small, efficient reasoning model with 24B parameters.",
      "sizes": [
        "24b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/magistral",
      "install": "ollama pull magistral"
    },
    {
      "id": "nomic-embed-text-v2-moe",
      "label": "nomic-embed-text-v2-moe",
      "description": "nomic-embed-text-v2-moe is a multilingual MoE text embedding model that excels at multilingual retrieval.",
      "sizes": [],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/nomic-embed-text-v2-moe",
      "install": "ollama pull nomic-embed-text-v2-moe"
    },
    {
      "id": "mistral-openorca",
      "label": "mistral-openorca",
      "description": "Mistral OpenOrca is a 7 billion parameter model, fine-tuned on top of the Mistral 7B model using the OpenOrca dataset.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/mistral-openorca",
      "install": "ollama pull mistral-openorca"
    },
    {
      "id": "openchat",
      "label": "openchat",
      "description": "A family of open-source models trained on a wide variety of data, surpassing ChatGPT on various benchmarks. Updated to version 3.5-0106.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/openchat",
      "install": "ollama pull openchat"
    },
    {
      "id": "codestral",
      "label": "codestral",
      "description": "Codestral is Mistral AI’s first-ever code model designed for code generation tasks.",
      "sizes": [
        "22b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/codestral",
      "install": "ollama pull codestral"
    },
    {
      "id": "mistral-large",
      "label": "mistral-large",
      "description": "Mistral Large 2 is Mistral's new flagship model that is significantly more capable in code generation, mathematics, and reasoning with 128k context window and support for dozens of languages.",
      "sizes": [
        "123b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/mistral-large",
      "install": "ollama pull mistral-large"
    },
    {
      "id": "wizard-vicuna-uncensored",
      "label": "wizard-vicuna-uncensored",
      "description": "Wizard Vicuna Uncensored is a 7B, 13B, and 30B parameter model based on Llama 2 uncensored by Eric Hartford.",
      "sizes": [
        "7b",
        "13b",
        "30b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/wizard-vicuna-uncensored",
      "install": "ollama pull wizard-vicuna-uncensored"
    },
    {
      "id": "vicuna",
      "label": "vicuna",
      "description": "General use chat model based on Llama and Llama 2 with 2K to 16K context sizes.",
      "sizes": [
        "7b",
        "13b",
        "33b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/vicuna",
      "install": "ollama pull vicuna"
    },
    {
      "id": "deepseek-llm",
      "label": "deepseek-llm",
      "description": "An advanced language model crafted with 2 trillion bilingual tokens.",
      "sizes": [
        "7b",
        "67b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/deepseek-llm",
      "install": "ollama pull deepseek-llm"
    },
    {
      "id": "zephyr",
      "label": "zephyr",
      "description": "Zephyr is a series of fine-tuned versions of the Mistral and Mixtral models that are trained to act as helpful assistants.",
      "sizes": [
        "7b",
        "141b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/zephyr",
      "install": "ollama pull zephyr"
    },
    {
      "id": "lfm2.5-thinking",
      "label": "lfm2.5-thinking",
      "description": "LFM2.5 is a new family of hybrid models designed for on-device deployment.",
      "sizes": [
        "1.2b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/lfm2.5-thinking",
      "install": "ollama pull lfm2.5-thinking"
    },
    {
      "id": "glm4",
      "label": "glm4",
      "description": "A strong multi-lingual general language model with competitive performance to Llama 3.",
      "sizes": [
        "9b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/glm4",
      "install": "ollama pull glm4"
    },
    {
      "id": "nous-hermes",
      "label": "nous-hermes",
      "description": "General use models based on Llama and Llama 2 from Nous Research.",
      "sizes": [
        "7b",
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/nous-hermes",
      "install": "ollama pull nous-hermes"
    },
    {
      "id": "deepscaler",
      "label": "deepscaler",
      "description": "A fine-tuned version of Deepseek-R1-Distilled-Qwen-1.5B that surpasses the performance of OpenAI’s o1-preview with just 1.5B parameters on popular math evaluations.",
      "sizes": [
        "1.5b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/deepscaler",
      "install": "ollama pull deepscaler"
    },
    {
      "id": "deepseek-v2",
      "label": "deepseek-v2",
      "description": "A strong, economical, and efficient Mixture-of-Experts language model.",
      "sizes": [
        "16b",
        "236b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/deepseek-v2",
      "install": "ollama pull deepseek-v2"
    },
    {
      "id": "neural-chat",
      "label": "neural-chat",
      "description": "A fine-tuned model based on Mistral with good coverage of domain and language.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/neural-chat",
      "install": "ollama pull neural-chat"
    },
    {
      "id": "wizardlm2",
      "label": "wizardlm2",
      "description": "State of the art large language model from Microsoft AI with improved performance on complex chat, multilingual, reasoning and agent use cases.",
      "sizes": [
        "7b"
      ],
      "capabilities": [
        "8x22b"
      ],
      "url": "https://ollama.com/library/wizardlm2",
      "install": "ollama pull wizardlm2"
    },
    {
      "id": "falcon",
      "label": "falcon",
      "description": "A large language model built by the Technology Innovation Institute (TII) for use in summarization, text generation, and chat bots.",
      "sizes": [
        "7b",
        "40b",
        "180b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/falcon",
      "install": "ollama pull falcon"
    },
    {
      "id": "llama2-chinese",
      "label": "llama2-chinese",
      "description": "Llama 2 based model fine tuned to improve Chinese dialogue ability.",
      "sizes": [
        "7b",
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/llama2-chinese",
      "install": "ollama pull llama2-chinese"
    },
    {
      "id": "qwen2-math",
      "label": "qwen2-math",
      "description": "Qwen2 Math is a series of specialized math language models built upon the Qwen2 LLMs, which significantly outperforms the mathematical capabilities of open-source models and even closed-source models (e.g., GPT4o).",
      "sizes": [
        "1.5b",
        "7b",
        "72b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/qwen2-math",
      "install": "ollama pull qwen2-math"
    },
    {
      "id": "openthinker",
      "label": "openthinker",
      "description": "A fully open-source family of reasoning models built using a dataset derived by distilling DeepSeek-R1.",
      "sizes": [
        "7b",
        "32b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/openthinker",
      "install": "ollama pull openthinker"
    },
    {
      "id": "stablelm2",
      "label": "stablelm2",
      "description": "Stable LM 2 is a state-of-the-art 1.6B and 12B parameter language model trained on multilingual data in English, Spanish, German, Italian, French, Portuguese, and Dutch.",
      "sizes": [
        "1.6b",
        "12b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/stablelm2",
      "install": "ollama pull stablelm2"
    },
    {
      "id": "codeqwen",
      "label": "codeqwen",
      "description": "CodeQwen1.5 is a large language model pretrained on a large amount of code data.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/codeqwen",
      "install": "ollama pull codeqwen"
    },
    {
      "id": "granite3.3",
      "label": "granite3.3",
      "description": "IBM Granite 2B and 8B models are 128K context length language models that have been fine-tuned for improved reasoning and instruction-following capabilities.",
      "sizes": [
        "2b",
        "8b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/granite3.3",
      "install": "ollama pull granite3.3"
    },
    {
      "id": "yi-coder",
      "label": "yi-coder",
      "description": "Yi-Coder is a series of open-source code language models that delivers state-of-the-art coding performance with fewer than 10 billion parameters.",
      "sizes": [
        "1.5b",
        "9b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/yi-coder",
      "install": "ollama pull yi-coder"
    },
    {
      "id": "lfm2",
      "label": "lfm2",
      "description": "LFM2 is a family of hybrid models designed for on-device deployment. LFM2-24B-A2B is the largest model in the family, scaling the architecture to 24 billion parameters while keeping inference efficient.",
      "sizes": [
        "24b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/lfm2",
      "install": "ollama pull lfm2"
    },
    {
      "id": "aya",
      "label": "aya",
      "description": "Aya 23, released by Cohere, is a new family of state-of-the-art, multilingual models that support 23 languages.",
      "sizes": [
        "8b",
        "35b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/aya",
      "install": "ollama pull aya"
    },
    {
      "id": "llava-phi3",
      "label": "llava-phi3",
      "description": "A new small LLaVA model fine-tuned from Phi 3 Mini.",
      "sizes": [
        "3.8b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/llava-phi3",
      "install": "ollama pull llava-phi3"
    },
    {
      "id": "nous-hermes2",
      "label": "nous-hermes2",
      "description": "The powerful family of models by Nous Research that excels at scientific discussion and coding tasks.",
      "sizes": [
        "10.7b",
        "34b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/nous-hermes2",
      "install": "ollama pull nous-hermes2"
    },
    {
      "id": "stable-code",
      "label": "stable-code",
      "description": "Stable Code 3B is a coding model with instruct and code completion variants on par with models such as Code Llama 7B that are 2.5x larger.",
      "sizes": [
        "3b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/stable-code",
      "install": "ollama pull stable-code"
    },
    {
      "id": "devstral",
      "label": "devstral",
      "description": "Devstral: the best open source model for coding agents",
      "sizes": [
        "24b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/devstral",
      "install": "ollama pull devstral"
    },
    {
      "id": "wizard-math",
      "label": "wizard-math",
      "description": "Model focused on math and logic problems",
      "sizes": [
        "7b",
        "13b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/wizard-math",
      "install": "ollama pull wizard-math"
    },
    {
      "id": "internlm2",
      "label": "internlm2",
      "description": "InternLM2.5 is a 7B parameter model tailored for practical scenarios with outstanding reasoning capability.",
      "sizes": [
        "1m",
        "1.8b",
        "7b",
        "20b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/internlm2",
      "install": "ollama pull internlm2"
    },
    {
      "id": "llama-guard3",
      "label": "llama-guard3",
      "description": "Llama Guard 3 is a series of models fine-tuned for content safety classification of LLM inputs and responses.",
      "sizes": [
        "1b",
        "8b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/llama-guard3",
      "install": "ollama pull llama-guard3"
    },
    {
      "id": "granite3.1-dense",
      "label": "granite3.1-dense",
      "description": "The IBM Granite 2B and 8B models are text-only dense LLMs trained on over 12 trillion tokens of data, demonstrated significant improvements over their predecessors in performance and speed in IBM’s initial testing.",
      "sizes": [
        "2b",
        "8b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/granite3.1-dense",
      "install": "ollama pull granite3.1-dense"
    },
    {
      "id": "aya-expanse",
      "label": "aya-expanse",
      "description": "Cohere For AI's language models trained to perform well across 23 different languages.",
      "sizes": [
        "8b",
        "32b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/aya-expanse",
      "install": "ollama pull aya-expanse"
    },
    {
      "id": "xwinlm",
      "label": "xwinlm",
      "description": "Conversational model based on Llama 2 that performs competitively on various benchmarks.",
      "sizes": [
        "7b",
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/xwinlm",
      "install": "ollama pull xwinlm"
    },
    {
      "id": "llama3-chatqa",
      "label": "llama3-chatqa",
      "description": "A model from NVIDIA based on Llama 3 that excels at conversational question answering (QA) and retrieval-augmented generation (RAG).",
      "sizes": [
        "8b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/llama3-chatqa",
      "install": "ollama pull llama3-chatqa"
    },
    {
      "id": "granite3-dense",
      "label": "granite3-dense",
      "description": "The IBM Granite 2B and 8B models are designed to support tool-based use cases and support for retrieval augmented generation (RAG), streamlining code generation, translation and bug fixing.",
      "sizes": [
        "2b",
        "8b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/granite3-dense",
      "install": "ollama pull granite3-dense"
    },
    {
      "id": "phi3.5",
      "label": "phi3.5",
      "description": "A lightweight AI model with 3.8 billion parameters with performance overtaking similarly and larger sized models.",
      "sizes": [
        "3.8b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/phi3.5",
      "install": "ollama pull phi3.5"
    },
    {
      "id": "dolphincoder",
      "label": "dolphincoder",
      "description": "A 7B and 15B uncensored variant of the Dolphin model family that excels at coding, based on StarCoder2.",
      "sizes": [
        "7b",
        "15b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/dolphincoder",
      "install": "ollama pull dolphincoder"
    },
    {
      "id": "samantha-mistral",
      "label": "samantha-mistral",
      "description": "A companion assistant trained in philosophy, psychology, and personal relationships. Based on Mistral.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/samantha-mistral",
      "install": "ollama pull samantha-mistral"
    },
    {
      "id": "yarn-llama2",
      "label": "yarn-llama2",
      "description": "An extension of Llama 2 that supports a context of up to 128k tokens.",
      "sizes": [
        "7b",
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/yarn-llama2",
      "install": "ollama pull yarn-llama2"
    },
    {
      "id": "llama3-gradient",
      "label": "llama3-gradient",
      "description": "This model extends LLama-3 8B's context length from 8k to over 1m tokens.",
      "sizes": [
        "8b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/llama3-gradient",
      "install": "ollama pull llama3-gradient"
    },
    {
      "id": "llama3-groq-tool-use",
      "label": "llama3-groq-tool-use",
      "description": "A series of models from Groq that represent a significant advancement in open-source AI capabilities for tool use/function calling.",
      "sizes": [
        "8b",
        "70b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/llama3-groq-tool-use",
      "install": "ollama pull llama3-groq-tool-use"
    },
    {
      "id": "granite3.2-vision",
      "label": "granite3.2-vision",
      "description": "A compact and efficient vision-language model, specifically designed for visual document understanding, enabling automated content extraction from tables, charts, infographics, plots, diagrams, and more.",
      "sizes": [
        "2b"
      ],
      "capabilities": [
        "vision",
        "tools"
      ],
      "url": "https://ollama.com/library/granite3.2-vision",
      "install": "ollama pull granite3.2-vision"
    },
    {
      "id": "phind-codellama",
      "label": "phind-codellama",
      "description": "Code generation model based on Code Llama.",
      "sizes": [
        "34b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/phind-codellama",
      "install": "ollama pull phind-codellama"
    },
    {
      "id": "starling-lm",
      "label": "starling-lm",
      "description": "Starling is a large language model trained by reinforcement learning from AI feedback focused on improving chatbot helpfulness.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/starling-lm",
      "install": "ollama pull starling-lm"
    },
    {
      "id": "solar",
      "label": "solar",
      "description": "A compact, yet powerful 10.7B large language model designed for single-turn conversation.",
      "sizes": [
        "10.7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/solar",
      "install": "ollama pull solar"
    },
    {
      "id": "stable-beluga",
      "label": "stable-beluga",
      "description": "Llama 2 based model fine tuned on an Orca-style dataset. Originally called Free Willy.",
      "sizes": [
        "7b",
        "13b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/stable-beluga",
      "install": "ollama pull stable-beluga"
    },
    {
      "id": "granite3-moe",
      "label": "granite3-moe",
      "description": "The IBM Granite 1B and 3B models are the first mixture of experts (MoE) Granite models from IBM designed for low latency usage.",
      "sizes": [
        "1b",
        "3b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/granite3-moe",
      "install": "ollama pull granite3-moe"
    },
    {
      "id": "nemotron-3-nano",
      "label": "nemotron-3-nano",
      "description": "Nemotron-3-Nano is a new Standard for Efficient, Open, and Intelligent Agentic Models, now updated with a 4B parameter count model.",
      "sizes": [
        "4b",
        "30b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/nemotron-3-nano",
      "install": "ollama pull nemotron-3-nano"
    },
    {
      "id": "shieldgemma",
      "label": "shieldgemma",
      "description": "ShieldGemma is set of instruction tuned models for evaluating the safety of text prompt input and text output responses against a set of defined safety policies.",
      "sizes": [
        "2b",
        "9b",
        "27b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/shieldgemma",
      "install": "ollama pull shieldgemma"
    },
    {
      "id": "orca2",
      "label": "orca2",
      "description": "Orca 2 is built by Microsoft research, and are a fine-tuned version of Meta's Llama 2 models. The model is designed to excel particularly in reasoning.",
      "sizes": [
        "7b",
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/orca2",
      "install": "ollama pull orca2"
    },
    {
      "id": "deepcoder",
      "label": "deepcoder",
      "description": "DeepCoder is a fully open-Source 14B coder model at O3-mini level, with a 1.5B version also available.",
      "sizes": [
        "1.5b",
        "14b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/deepcoder",
      "install": "ollama pull deepcoder"
    },
    {
      "id": "wizardlm",
      "label": "wizardlm",
      "description": "General use model based on Llama 2.",
      "sizes": [],
      "capabilities": [],
      "url": "https://ollama.com/library/wizardlm",
      "install": "ollama pull wizardlm"
    },
    {
      "id": "reader-lm",
      "label": "reader-lm",
      "description": "A series of models that convert HTML content to Markdown content, which is useful for content conversion tasks.",
      "sizes": [
        "0.5b",
        "1.5b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/reader-lm",
      "install": "ollama pull reader-lm"
    },
    {
      "id": "paraphrase-multilingual",
      "label": "paraphrase-multilingual",
      "description": "Sentence-transformers model that can be used for tasks like clustering or semantic search.",
      "sizes": [
        "278m"
      ],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/paraphrase-multilingual",
      "install": "ollama pull paraphrase-multilingual"
    },
    {
      "id": "llama-pro",
      "label": "llama-pro",
      "description": "An expansion of Llama 2 that specializes in integrating both general language understanding and domain-specific knowledge, particularly in programming and mathematics.",
      "sizes": [],
      "capabilities": [],
      "url": "https://ollama.com/library/llama-pro",
      "install": "ollama pull llama-pro"
    },
    {
      "id": "yarn-mistral",
      "label": "yarn-mistral",
      "description": "An extension of Mistral to support context windows of 64K or 128K.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/yarn-mistral",
      "install": "ollama pull yarn-mistral"
    },
    {
      "id": "nexusraven",
      "label": "nexusraven",
      "description": "Nexus Raven is a 13B instruction tuned model for function calling tasks.",
      "sizes": [
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/nexusraven",
      "install": "ollama pull nexusraven"
    },
    {
      "id": "meditron",
      "label": "meditron",
      "description": "Open-source medical large language model adapted from Llama 2 to the medical domain.",
      "sizes": [
        "7b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/meditron",
      "install": "ollama pull meditron"
    },
    {
      "id": "bakllava",
      "label": "bakllava",
      "description": "BakLLaVA is a multimodal model consisting of the Mistral 7B base model augmented with the LLaVA architecture.",
      "sizes": [
        "7b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/bakllava",
      "install": "ollama pull bakllava"
    },
    {
      "id": "mistral-small3.1",
      "label": "mistral-small3.1",
      "description": "Building upon Mistral Small 3, Mistral Small 3.1 (2503) adds state-of-the-art vision understanding and enhances long context capabilities up to 128k tokens without compromising text performance.",
      "sizes": [
        "24b"
      ],
      "capabilities": [
        "vision",
        "tools"
      ],
      "url": "https://ollama.com/library/mistral-small3.1",
      "install": "ollama pull mistral-small3.1"
    },
    {
      "id": "nemotron-3.5-lightning",
      "label": "nemotron-3.5-lightning",
      "description": "NVIDIA Nemotron 3.5 Lightning is an open 30B mixture-of-experts (MoE) model with 3B active parameters built for always-on agents.",
      "sizes": [
        "30b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/nemotron-3.5-lightning",
      "install": "ollama pull nemotron-3.5-lightning"
    },
    {
      "id": "command-r-plus",
      "label": "command-r-plus",
      "description": "Command R&#43; is a powerful, scalable large language model purpose-built to excel at real-world enterprise use cases.",
      "sizes": [
        "104b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/command-r-plus",
      "install": "ollama pull command-r-plus"
    },
    {
      "id": "exaone-deep",
      "label": "exaone-deep",
      "description": "EXAONE Deep exhibits superior capabilities in various reasoning tasks including math and coding benchmarks, ranging from 2.4B to 32B parameters developed and released by LG AI Research.",
      "sizes": [
        "2.4b",
        "7.8b",
        "32b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/exaone-deep",
      "install": "ollama pull exaone-deep"
    },
    {
      "id": "ornith",
      "label": "ornith",
      "description": "A self-improving family of open-source models for agentic coding",
      "sizes": [
        "9b",
        "35b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/ornith",
      "install": "ollama pull ornith"
    },
    {
      "id": "tinydolphin",
      "label": "tinydolphin",
      "description": "An experimental 1.1B parameter model trained on the new Dolphin 2.8 dataset by Eric Hartford and based on TinyLlama.",
      "sizes": [
        "1.1b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/tinydolphin",
      "install": "ollama pull tinydolphin"
    },
    {
      "id": "medllama2",
      "label": "medllama2",
      "description": "Fine-tuned Llama 2 model to answer medical questions based on an open source medical dataset.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/medllama2",
      "install": "ollama pull medllama2"
    },
    {
      "id": "deepseek-v3.1",
      "label": "deepseek-v3.1",
      "description": "DeepSeek-V3.1-Terminus is a hybrid model that supports both thinking mode and non-thinking mode.",
      "sizes": [
        "671b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/deepseek-v3.1",
      "install": "ollama pull deepseek-v3.1"
    },
    {
      "id": "nemotron-mini",
      "label": "nemotron-mini",
      "description": "A commercial-friendly small language model by NVIDIA optimized for roleplay, RAG QA, and function calling.",
      "sizes": [
        "4b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/nemotron-mini",
      "install": "ollama pull nemotron-mini"
    },
    {
      "id": "nemotron3",
      "label": "nemotron3",
      "description": "NVIDIA Nemotron 3 Nano Omni is a multimodal large language model that unifies video, audio, image, and text understanding to support enterprise-grade Q&A, summarization, transcription, and document intelligence workflows.",
      "sizes": [
        "33b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "audio"
      ],
      "url": "https://ollama.com/library/nemotron3",
      "install": "ollama pull nemotron3"
    },
    {
      "id": "codegeex4",
      "label": "codegeex4",
      "description": "A versatile model for AI software development scenarios, including code completion.",
      "sizes": [
        "9b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/codegeex4",
      "install": "ollama pull codegeex4"
    },
    {
      "id": "deepseek-ocr",
      "label": "deepseek-ocr",
      "description": "DeepSeek-OCR is a vision-language model that can perform token-efficient OCR.",
      "sizes": [
        "3b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/deepseek-ocr",
      "install": "ollama pull deepseek-ocr"
    },
    {
      "id": "opencoder",
      "label": "opencoder",
      "description": "OpenCoder is an open and reproducible code LLM family which includes 1.5B and 8B models, supporting chat in English and Chinese languages.",
      "sizes": [
        "1.5b",
        "8b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/opencoder",
      "install": "ollama pull opencoder"
    },
    {
      "id": "wizardlm-uncensored",
      "label": "wizardlm-uncensored",
      "description": "Uncensored version of Wizard LM model",
      "sizes": [
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/wizardlm-uncensored",
      "install": "ollama pull wizardlm-uncensored"
    },
    {
      "id": "nemotron",
      "label": "nemotron",
      "description": "Llama-3.1-Nemotron-70B-Instruct is a large language model customized by NVIDIA to improve the helpfulness of LLM generated responses to user queries.",
      "sizes": [
        "70b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/nemotron",
      "install": "ollama pull nemotron"
    },
    {
      "id": "reflection",
      "label": "reflection",
      "description": "A high-performing model trained with a new technique called Reflection-tuning that teaches a LLM to detect mistakes in its reasoning and correct course.",
      "sizes": [
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/reflection",
      "install": "ollama pull reflection"
    },
    {
      "id": "granite4.1",
      "label": "granite4.1",
      "description": "IBM Granite Models are a family of enterprise-ready, open foundation models that support multilingual capabilities, coding, retrieval-augmented generation (RAG), tool use, and structured JSON output. Released under Apache 2.0 license.",
      "sizes": [
        "3b",
        "8b",
        "30b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/granite4.1",
      "install": "ollama pull granite4.1"
    },
    {
      "id": "codeup",
      "label": "codeup",
      "description": "Great code generation model based on Llama2.",
      "sizes": [
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/codeup",
      "install": "ollama pull codeup"
    },
    {
      "id": "athene-v2",
      "label": "athene-v2",
      "description": "Athene-V2 is a 72B parameter model which excels at code completion, mathematics, and log extraction tasks.",
      "sizes": [
        "72b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/athene-v2",
      "install": "ollama pull athene-v2"
    },
    {
      "id": "nous-hermes2-mixtral",
      "label": "nous-hermes2-mixtral",
      "description": "The Nous Hermes 2 model from Nous Research, now trained over Mixtral.",
      "sizes": [],
      "capabilities": [
        "8x7b"
      ],
      "url": "https://ollama.com/library/nous-hermes2-mixtral",
      "install": "ollama pull nous-hermes2-mixtral"
    },
    {
      "id": "qwen3-next",
      "label": "qwen3-next",
      "description": "The first installment in the Qwen3-Next series with strong performance in terms of both parameter efficiency and inference speed.",
      "sizes": [
        "80b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/qwen3-next",
      "install": "ollama pull qwen3-next"
    },
    {
      "id": "exaone3.5",
      "label": "exaone3.5",
      "description": "EXAONE 3.5 is a collection of instruction-tuned bilingual (English and Korean) generative models ranging from 2.4B to 32B parameters, developed and released by LG AI Research.",
      "sizes": [
        "2.4b",
        "7.8b",
        "32b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/exaone3.5",
      "install": "ollama pull exaone3.5"
    },
    {
      "id": "megadolphin",
      "label": "megadolphin",
      "description": "MegaDolphin-2.2-120b is a transformation of Dolphin-2.2-70b created by interleaving the model with itself.",
      "sizes": [
        "120b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/megadolphin",
      "install": "ollama pull megadolphin"
    },
    {
      "id": "everythinglm",
      "label": "everythinglm",
      "description": "Uncensored Llama2 based model with support for a 16K context window.",
      "sizes": [
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/everythinglm",
      "install": "ollama pull everythinglm"
    },
    {
      "id": "nuextract",
      "label": "nuextract",
      "description": "A 3.8B model fine-tuned on a private high-quality synthetic dataset for information extraction, based on Phi-3.",
      "sizes": [
        "3.8b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/nuextract",
      "install": "ollama pull nuextract"
    },
    {
      "id": "solar-pro",
      "label": "solar-pro",
      "description": "Solar Pro Preview: an advanced large language model (LLM) with 22 billion parameters designed to fit into a single GPU",
      "sizes": [
        "22b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/solar-pro",
      "install": "ollama pull solar-pro"
    },
    {
      "id": "mathstral",
      "label": "mathstral",
      "description": "MathΣtral: a 7B model designed for math reasoning and scientific discovery by Mistral AI.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/mathstral",
      "install": "ollama pull mathstral"
    },
    {
      "id": "magicoder",
      "label": "magicoder",
      "description": "🎩 Magicoder is a family of 7B parameter models trained on 75K synthetic instruction data using OSS-Instruct, a novel approach to enlightening LLMs with open-source code snippets.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/magicoder",
      "install": "ollama pull magicoder"
    },
    {
      "id": "notus",
      "label": "notus",
      "description": "A 7B chat model fine-tuned with high-quality data and based on Zephyr.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/notus",
      "install": "ollama pull notus"
    },
    {
      "id": "falcon2",
      "label": "falcon2",
      "description": "Falcon2 is an 11B parameters causal decoder-only model built by TII and trained over 5T tokens.",
      "sizes": [
        "11b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/falcon2",
      "install": "ollama pull falcon2"
    },
    {
      "id": "notux",
      "label": "notux",
      "description": "A top-performing mixture of experts model, fine-tuned with high-quality data.",
      "sizes": [],
      "capabilities": [
        "8x7b"
      ],
      "url": "https://ollama.com/library/notux",
      "install": "ollama pull notux"
    },
    {
      "id": "stablelm-zephyr",
      "label": "stablelm-zephyr",
      "description": "A lightweight chat model allowing accurate, and responsive output without requiring high-end hardware.",
      "sizes": [
        "3b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/stablelm-zephyr",
      "install": "ollama pull stablelm-zephyr"
    },
    {
      "id": "bespoke-minicheck",
      "label": "bespoke-minicheck",
      "description": "A state-of-the-art fact-checking model developed by Bespoke Labs.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/bespoke-minicheck",
      "install": "ollama pull bespoke-minicheck"
    },
    {
      "id": "duckdb-nsql",
      "label": "duckdb-nsql",
      "description": "7B parameter text-to-SQL model made by MotherDuck and Numbers Station.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/duckdb-nsql",
      "install": "ollama pull duckdb-nsql"
    },
    {
      "id": "firefunction-v2",
      "label": "firefunction-v2",
      "description": "An open weights function calling model based on Llama 3, competitive with GPT-4o function calling capabilities.",
      "sizes": [
        "70b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/firefunction-v2",
      "install": "ollama pull firefunction-v2"
    },
    {
      "id": "mistrallite",
      "label": "mistrallite",
      "description": "MistralLite is a fine-tuned model based on Mistral with enhanced capabilities of processing long contexts.",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/mistrallite",
      "install": "ollama pull mistrallite"
    },
    {
      "id": "minimax-m3",
      "label": "minimax-m3",
      "description": "MiniMax M3: Coding & Agentic Frontier. 1M context window. Native Multimodality.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/minimax-m3",
      "install": "ollama pull minimax-m3"
    },
    {
      "id": "wizard-vicuna",
      "label": "wizard-vicuna",
      "description": "Wizard Vicuna is a 13B parameter model based on Llama 2 trained by MelodysDreamj.",
      "sizes": [
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/wizard-vicuna",
      "install": "ollama pull wizard-vicuna"
    },
    {
      "id": "open-orca-platypus2",
      "label": "open-orca-platypus2",
      "description": "Merge of the Open Orca OpenChat model and the Garage-bAInd Platypus 2 model. Designed for chat and code generation.",
      "sizes": [
        "13b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/open-orca-platypus2",
      "install": "ollama pull open-orca-platypus2"
    },
    {
      "id": "rnj-1",
      "label": "rnj-1",
      "description": "Rnj-1 is a family of 8B parameter open-weight, dense models trained from scratch by Essential AI, optimized for code and STEM with capabilities on par with SOTA open-weight models.",
      "sizes": [
        "8b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/rnj-1",
      "install": "ollama pull rnj-1"
    },
    {
      "id": "codebooga",
      "label": "codebooga",
      "description": "A high-performing code instruct model created by merging two existing code models.",
      "sizes": [
        "34b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/codebooga",
      "install": "ollama pull codebooga"
    },
    {
      "id": "kimi-k2.6",
      "label": "kimi-k2.6",
      "description": "Kimi K2.6 is an open-source, native multimodal agentic model that advances practical capabilities in long-horizon coding, coding-driven design, proactive autonomous execution, and swarm-based task orchestration.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/kimi-k2.6",
      "install": "ollama pull kimi-k2.6"
    },
    {
      "id": "deepseek-v4-pro",
      "label": "deepseek-v4-pro",
      "description": "DeepSeek-V4-Pro is a frontier Mixture-of-Experts model with a large context window and three reasoning modes.",
      "sizes": [],
      "capabilities": [
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/deepseek-v4-pro",
      "install": "ollama pull deepseek-v4-pro"
    },
    {
      "id": "goliath",
      "label": "goliath",
      "description": "A language model created by combining two fine-tuned Llama 2 70B models into one.",
      "sizes": [],
      "capabilities": [],
      "url": "https://ollama.com/library/goliath",
      "install": "ollama pull goliath"
    },
    {
      "id": "snowflake-arctic-embed2",
      "label": "snowflake-arctic-embed2",
      "description": "Snowflake's frontier embedding model. Arctic Embed 2.0 adds multilingual support without sacrificing English performance or scalability.",
      "sizes": [
        "568m"
      ],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/snowflake-arctic-embed2",
      "install": "ollama pull snowflake-arctic-embed2"
    },
    {
      "id": "olmo-3",
      "label": "olmo-3",
      "description": "Olmo is a series of Open language models designed to enable the science of language models. These models are pre-trained on the Dolma 3 dataset and post-trained on the Dolci datasets.",
      "sizes": [
        "7b",
        "32b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/olmo-3",
      "install": "ollama pull olmo-3"
    },
    {
      "id": "medgemma",
      "label": "medgemma",
      "description": "MedGemma is a collection of Gemma 3 variants that are trained for performance on medical text and image comprehension.",
      "sizes": [
        "4b",
        "27b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/medgemma",
      "install": "ollama pull medgemma"
    },
    {
      "id": "granite3.2",
      "label": "granite3.2",
      "description": "Granite-3.2 is a family of long-context AI models from IBM Granite fine-tuned for thinking capabilities.",
      "sizes": [
        "2b",
        "8b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/granite3.2",
      "install": "ollama pull granite3.2"
    },
    {
      "id": "ornith-1.5",
      "label": "ornith-1.5",
      "description": "Chirp Chirp! 🐦 We are introducing Ornith-1.5, a major step toward building foundation models through end-to-end self-improvement.",
      "sizes": [
        "9b",
        "35b",
        "122b",
        "397b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/ornith-1.5",
      "install": "ollama pull ornith-1.5"
    },
    {
      "id": "sailor2",
      "label": "sailor2",
      "description": "Sailor2 are multilingual language models made for South-East Asia. Available in 1B, 8B, and 20B parameter sizes.",
      "sizes": [
        "1b",
        "8b",
        "20b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/sailor2",
      "install": "ollama pull sailor2"
    },
    {
      "id": "mistral-medium-3.5",
      "label": "mistral-medium-3.5",
      "description": "Mistral Medium 3.5 is the first flagship model of Mistral AI that merged instruction-following, reasoning, and coding in a single set of 128B weights.",
      "sizes": [
        "128b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/mistral-medium-3.5",
      "install": "ollama pull mistral-medium-3.5"
    },
    {
      "id": "r1-1776",
      "label": "r1-1776",
      "description": "A version of the DeepSeek-R1 model that has been post trained to provide unbiased, accurate, and factual information by Perplexity.",
      "sizes": [
        "70b",
        "671b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/r1-1776",
      "install": "ollama pull r1-1776"
    },
    {
      "id": "tulu3",
      "label": "tulu3",
      "description": "Tülu 3 is a leading instruction following model family, offering fully open-source data, code, and recipes by the The Allen Institute for AI.",
      "sizes": [
        "8b",
        "70b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/tulu3",
      "install": "ollama pull tulu3"
    },
    {
      "id": "granite-embedding",
      "label": "granite-embedding",
      "description": "The IBM Granite Embedding 30M and 278M models models are text-only dense biencoder embedding models, with 30M available in English only and 278M serving multilingual use cases.",
      "sizes": [
        "30m",
        "278m"
      ],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/granite-embedding",
      "install": "ollama pull granite-embedding"
    },
    {
      "id": "muse-glimmer",
      "label": "muse-glimmer",
      "description": "Meta's latest open model built for always-on local agents. 30B parameters, licensed under Apache 2.0 and runs on a single GPU — tuned for tool use, long tasks, and failure recovery.",
      "sizes": [
        "30b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/muse-glimmer",
      "install": "ollama pull muse-glimmer"
    },
    {
      "id": "dbrx",
      "label": "dbrx",
      "description": "DBRX is an open, general-purpose LLM created by Databricks.",
      "sizes": [
        "132b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/dbrx",
      "install": "ollama pull dbrx"
    },
    {
      "id": "glm-5.2",
      "label": "glm-5.2",
      "description": "GLM-5.2 is Z.ai’s flagship model for the era of long-horizon tasks.",
      "sizes": [],
      "capabilities": [
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/glm-5.2",
      "install": "ollama pull glm-5.2"
    },
    {
      "id": "devstral-2",
      "label": "devstral-2",
      "description": "123B model that excels at using tools to explore codebases, editing multiple files and power software engineering agents.",
      "sizes": [
        "123b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/devstral-2",
      "install": "ollama pull devstral-2"
    },
    {
      "id": "granite3-guardian",
      "label": "granite3-guardian",
      "description": "The IBM Granite Guardian 3.0 2B and 8B models are designed to detect risks in prompts and/or responses.",
      "sizes": [
        "2b",
        "8b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/granite3-guardian",
      "install": "ollama pull granite3-guardian"
    },
    {
      "id": "command-r7b",
      "label": "command-r7b",
      "description": "The smallest model in Cohere's R series delivers top-tier speed, efficiency, and quality to build powerful AI applications on commodity GPUs and edge devices.",
      "sizes": [
        "7b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/command-r7b",
      "install": "ollama pull command-r7b"
    },
    {
      "id": "alfred",
      "label": "alfred",
      "description": "A robust conversational model designed to be used for both chat and instruct use cases.",
      "sizes": [
        "40b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/alfred",
      "install": "ollama pull alfred"
    },
    {
      "id": "lfm2.5",
      "label": "lfm2.5",
      "description": "LFM2.5-8B-A1B, an edge model built for fast, reliable tool calling on consumer hardware.",
      "sizes": [
        "8b",
        "24b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/lfm2.5",
      "install": "ollama pull lfm2.5"
    },
    {
      "id": "phi4-mini-reasoning",
      "label": "phi4-mini-reasoning",
      "description": "Phi 4 mini reasoning is a lightweight open model that balances efficiency with advanced reasoning ability.",
      "sizes": [
        "3.8b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/phi4-mini-reasoning",
      "install": "ollama pull phi4-mini-reasoning"
    },
    {
      "id": "olmo-3.1",
      "label": "olmo-3.1",
      "description": "Olmo is a series of Open language models designed to enable the science of language models. These models are pre-trained on the Dolma 3 dataset and post-trained on the Dolci datasets.",
      "sizes": [
        "32b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/olmo-3.1",
      "install": "ollama pull olmo-3.1"
    },
    {
      "id": "deepseek-v2.5",
      "label": "deepseek-v2.5",
      "description": "An upgraded version of DeekSeek-V2 that integrates the general and coding abilities of both DeepSeek-V2-Chat and DeepSeek-Coder-V2-Instruct.",
      "sizes": [
        "236b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/deepseek-v2.5",
      "install": "ollama pull deepseek-v2.5"
    },
    {
      "id": "bge-large",
      "label": "bge-large",
      "description": "Embedding model from BAAI mapping texts to vectors.",
      "sizes": [
        "335m"
      ],
      "capabilities": [
        "embedding"
      ],
      "url": "https://ollama.com/library/bge-large",
      "install": "ollama pull bge-large"
    },
    {
      "id": "granite4.2",
      "label": "granite4.2",
      "description": "IBM Granite Models are a family of enterprise-ready, open foundation models that support multilingual capabilities, coding, retrieval-augmented generation (RAG), tool use, thinking and structured JSON output. Released under Apache 2.0 license.",
      "sizes": [
        "3b",
        "8b",
        "30b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/granite4.2",
      "install": "ollama pull granite4.2"
    },
    {
      "id": "smallthinker",
      "label": "smallthinker",
      "description": "A new small reasoning model fine-tuned from the Qwen 2.5 3B Instruct model.",
      "sizes": [
        "3b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/smallthinker",
      "install": "ollama pull smallthinker"
    },
    {
      "id": "kimi-k2.7-code",
      "label": "kimi-k2.7-code",
      "description": "Kimi K2.7 Code is Moonshot AI's coding-focused agentic model built upon Kimi K2.6, with substantial improvements on real-world long-horizon coding tasks and roughly 30% lower thinking-token usage.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/kimi-k2.7-code",
      "install": "ollama pull kimi-k2.7-code"
    },
    {
      "id": "qwen3.8-flash-next",
      "label": "qwen3.8-flash-next",
      "description": "This experimental preview of the architecture that will underpin Qwen4.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/qwen3.8-flash-next",
      "install": "ollama pull qwen3.8-flash-next"
    },
    {
      "id": "command-a",
      "label": "command-a",
      "description": "111 billion parameter model optimized for demanding enterprises that require fast, secure, and high-quality AI",
      "sizes": [
        "111b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/command-a",
      "install": "ollama pull command-a"
    },
    {
      "id": "glm-5.3",
      "label": "glm-5.3",
      "description": "Z.ai's flagship model and the most capable open-weights model for coding, with major gains on long-horizon agentic tasks.",
      "sizes": [],
      "capabilities": [
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/glm-5.3",
      "install": "ollama pull glm-5.3"
    },
    {
      "id": "cogito-2.1",
      "label": "cogito-2.1",
      "description": "The Cogito v2.1 LLMs are instruction tuned generative models. All models are released under MIT license for commercial use.",
      "sizes": [
        "671b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/cogito-2.1",
      "install": "ollama pull cogito-2.1"
    },
    {
      "id": "marco-o1",
      "label": "marco-o1",
      "description": "An open large reasoning model for real-world solutions by the Alibaba International Digital Commerce Group (AIDC-AI).",
      "sizes": [
        "7b"
      ],
      "capabilities": [],
      "url": "https://ollama.com/library/marco-o1",
      "install": "ollama pull marco-o1"
    },
    {
      "id": "deepseek-v4.1-flash",
      "label": "deepseek-v4.1-flash",
      "description": "Designed for greater capability, faster inference, higher throughput, and scaling to larger models.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/deepseek-v4.1-flash",
      "install": "ollama pull deepseek-v4.1-flash"
    },
    {
      "id": "medgemma1.5",
      "label": "medgemma1.5",
      "description": "MedGemma 1.5 4B is an updated version of the MedGemma 4B model.",
      "sizes": [
        "4b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/medgemma1.5",
      "install": "ollama pull medgemma1.5"
    },
    {
      "id": "glm-5.3-flash",
      "label": "glm-5.3-flash",
      "description": "Z.ai's first natively multimodal model, approaching Claude Opus 4.8 on coding and agentic benchmarks with just 18B active parameters.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/glm-5.3-flash",
      "install": "ollama pull glm-5.3-flash"
    },
    {
      "id": "command-r7b-arabic",
      "label": "command-r7b-arabic",
      "description": "A new state-of-the-art version of the lightweight Command R7B model that excels in advanced Arabic language capabilities for enterprises in the Middle East and Northern Africa.",
      "sizes": [
        "7b"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/command-r7b-arabic",
      "install": "ollama pull command-r7b-arabic"
    },
    {
      "id": "functiongemma",
      "label": "functiongemma",
      "description": "FunctionGemma is a specialized version of Google's Gemma 3 270M model fine-tuned explicitly for function calling.",
      "sizes": [
        "270m"
      ],
      "capabilities": [
        "tools"
      ],
      "url": "https://ollama.com/library/functiongemma",
      "install": "ollama pull functiongemma"
    },
    {
      "id": "laguna-s-2.1",
      "label": "laguna-s-2.1",
      "description": "Our most capable model to date, designed for long-horizon work. 70.2% on Terminal-Bench 2.1 at 118B-A8B.",
      "sizes": [],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/laguna-s-2.1",
      "install": "ollama pull laguna-s-2.1"
    },
    {
      "id": "gpt-oss-safeguard",
      "label": "gpt-oss-safeguard",
      "description": "gpt-oss-safeguard-20b and gpt-oss-safeguard-120b are safety reasoning models built-upon gpt-oss",
      "sizes": [
        "20b",
        "120b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/gpt-oss-safeguard",
      "install": "ollama pull gpt-oss-safeguard"
    },
    {
      "id": "nemotron-cascade-2",
      "label": "nemotron-cascade-2",
      "description": "An open 30B MoE model from NVIDIA with 3B activated parameters that delivers strong reasoning and agentic capabilities.",
      "sizes": [
        "30b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/nemotron-cascade-2",
      "install": "ollama pull nemotron-cascade-2"
    },
    {
      "id": "bonsai-2",
      "label": "bonsai-2",
      "description": "Bonsai 2 from Prism ML is a family of vision models built for ternary weights – smaller models with close to full-precision quality.",
      "sizes": [
        "2b",
        "4b",
        "9b",
        "27b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/bonsai-2",
      "install": "ollama pull bonsai-2"
    },
    {
      "id": "tev1",
      "label": "tev1",
      "description": "A 4B decision model from Together AI for fast classification.",
      "sizes": [
        "0.8b",
        "4b"
      ],
      "capabilities": [
        "tools",
        "decision",
        "thinking"
      ],
      "url": "https://ollama.com/library/tev1",
      "install": "ollama pull tev1"
    },
    {
      "id": "embeddinggemma-2",
      "label": "embeddinggemma-2",
      "description": "EmbeddingGemma 2 is a multimodal embedding model from Google built on the Gemma 4 architecture.",
      "sizes": [
        "270m",
        "440m",
        "570m",
        "740m"
      ],
      "capabilities": [
        "vision",
        "embedding",
        "audio"
      ],
      "url": "https://ollama.com/library/embeddinggemma-2",
      "install": "ollama pull embeddinggemma-2"
    },
    {
      "id": "laguna-xs-2.1",
      "label": "laguna-xs-2.1",
      "description": "Laguna XS 2.1 is a 33B total parameter Mixture-of-Experts model with 3B activated parameters per token designed for agentic coding and long-horizon work on a local machine.",
      "sizes": [],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/laguna-xs-2.1",
      "install": "ollama pull laguna-xs-2.1"
    },
    {
      "id": "nemotron-3-ultra",
      "label": "nemotron-3-ultra",
      "description": "NVIDIA Nemotron 3 Ultra is built for high-throughput reasoning and long-running agent workflows.",
      "sizes": [],
      "capabilities": [
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/nemotron-3-ultra",
      "install": "ollama pull nemotron-3-ultra"
    },
    {
      "id": "phonon",
      "label": "phonon",
      "description": "Phonon is Gradium’s text-to-speech family. It turns written text into spoken audio as it generates, making it useful for reading assistant replies, narrating passages, and adding a voice to local apps.",
      "sizes": [],
      "capabilities": [],
      "url": "https://ollama.com/library/phonon",
      "install": "ollama pull phonon"
    },
    {
      "id": "kimi-k3",
      "label": "kimi-k3",
      "description": "Kimi K3 is an open-weight, native multimodal agentic model and our most capable model to date.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/kimi-k3",
      "install": "ollama pull kimi-k3"
    },
    {
      "id": "strands-decider",
      "label": "strands-decider",
      "description": "A 2B decision model from Strands Agents for fast classification, routing, and scoring.",
      "sizes": [
        "2b"
      ],
      "capabilities": [
        "decision"
      ],
      "url": "https://ollama.com/library/strands-decider",
      "install": "ollama pull strands-decider"
    },
    {
      "id": "clef",
      "label": "clef",
      "description": "Clef is a 27B multimodal model created by Cloudflare that turns a state and a schema of typed questions into decisions.",
      "sizes": [
        "27b"
      ],
      "capabilities": [
        "vision",
        "decision",
        "cloud"
      ],
      "url": "https://ollama.com/library/clef",
      "install": "ollama pull clef"
    },
    {
      "id": "minicpm-v4.6",
      "label": "minicpm-v4.6",
      "description": "A Pocket-Sized MLLM for Ultra-Efficient Image and Video Understanding on Your Phone",
      "sizes": [
        "1b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/minicpm-v4.6",
      "install": "ollama pull minicpm-v4.6"
    },
    {
      "id": "nimble",
      "label": "nimble",
      "description": "A 9B decision model from Bespoke Labs for fast, typed classification.",
      "sizes": [
        "9b"
      ],
      "capabilities": [
        "tools",
        "decision",
        "thinking"
      ],
      "url": "https://ollama.com/library/nimble",
      "install": "ollama pull nimble"
    },
    {
      "id": "north-mini-code-1.0",
      "label": "north-mini-code-1.0",
      "description": "North Mini Code is Cohere's first model for developers — a 30B Mixture-of-Experts model with 3B active parameters, built for agentic software engineering.",
      "sizes": [],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/north-mini-code-1.0",
      "install": "ollama pull north-mini-code-1.0"
    },
    {
      "id": "flint",
      "label": "flint",
      "description": "Flint is a 39.8 billion-parameter mixture-of-experts model from Recursive, built for coding agents.",
      "sizes": [
        "40b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/flint",
      "install": "ollama pull flint"
    },
    {
      "id": "d3",
      "label": "d3",
      "description": "Turn text, structured data, and images into scored choices, yes/no answers, and ratings.",
      "sizes": [
        "2b",
        "4b",
        "9b",
        "27b"
      ],
      "capabilities": [
        "vision",
        "decision"
      ],
      "url": "https://ollama.com/library/d3",
      "install": "ollama pull d3"
    },
    {
      "id": "clef-flash",
      "label": "clef-flash",
      "description": "Clef-Flash is a 9B multimodal model that turns a state and a schema of typed questions into decisions.",
      "sizes": [
        "9b"
      ],
      "capabilities": [
        "vision",
        "decision"
      ],
      "url": "https://ollama.com/library/clef-flash",
      "install": "ollama pull clef-flash"
    },
    {
      "id": "lfm3",
      "label": "lfm3",
      "description": "LFM3-1.7B is a tri-hybrid model for local agents and specialized applications.",
      "sizes": [
        "1.7b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/lfm3",
      "install": "ollama pull lfm3"
    },
    {
      "id": "beam",
      "label": "beam",
      "description": "Beam is an open model by Reflection designed to deliver strong performance for areas such as agentic tool use, coding and general reasoning with an emphasis on token efficient usage.",
      "sizes": [],
      "capabilities": [
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/beam",
      "install": "ollama pull beam"
    },
    {
      "id": "minicpm-v4.5",
      "label": "minicpm-v4.5",
      "description": "A GPT-4o Level MLLM for Single Image, Multi Image and High-FPS Video Understanding on Your Phone",
      "sizes": [
        "8b"
      ],
      "capabilities": [
        "vision"
      ],
      "url": "https://ollama.com/library/minicpm-v4.5",
      "install": "ollama pull minicpm-v4.5"
    },
    {
      "id": "mimo-v2.6",
      "label": "mimo-v2.6",
      "description": "MiMo-V2.6-Distill-Qwen-9B is Xiaomi MiMo’s 9B model for code and agent workflows",
      "sizes": [
        "9b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/mimo-v2.6",
      "install": "ollama pull mimo-v2.6"
    },
    {
      "id": "d1",
      "label": "d1",
      "description": "d1 is a 3B decision model from Liquid AI, built on LFM2.5-VL-3B",
      "sizes": [
        "2b",
        "3b"
      ],
      "capabilities": [
        "vision",
        "decision"
      ],
      "url": "https://ollama.com/library/d1",
      "install": "ollama pull d1"
    },
    {
      "id": "nemotron-3.5-bolt",
      "label": "nemotron-3.5-bolt",
      "description": "Nemotron 3.5 Bolt is NVIDIA’s compact vision-language model for local coding, reasoning, and tool use.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "audio",
        "e4b"
      ],
      "url": "https://ollama.com/library/nemotron-3.5-bolt",
      "install": "ollama pull nemotron-3.5-bolt"
    },
    {
      "id": "lfm3-omni",
      "label": "lfm3-omni",
      "description": "a single end-to-end model that understands text, images, video and audio, and answers in text and speech",
      "sizes": [
        "2b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "audio"
      ],
      "url": "https://ollama.com/library/lfm3-omni",
      "install": "ollama pull lfm3-omni"
    },
    {
      "id": "minimax-m3.1-flash",
      "label": "minimax-m3.1-flash",
      "description": "As a next-generation Flash model, M3.1 Flash delivers a generational leap in agentic tasks, coding, Computer Use and cybersecurity, while making online inference nearly three times faster and reducing usage costs by 70%.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/minimax-m3.1-flash",
      "install": "ollama pull minimax-m3.1-flash"
    },
    {
      "id": "d1-omni",
      "label": "d1-omni",
      "description": "d1-omni is a multimodal 600M decision model from Liquid AI",
      "sizes": [
        "600m",
        "2b"
      ],
      "capabilities": [
        "vision",
        "decision",
        "audio"
      ],
      "url": "https://ollama.com/library/d1-omni",
      "install": "ollama pull d1-omni"
    },
    {
      "id": "granite4.1-guardian",
      "label": "granite4.1-guardian",
      "description": "Granite Guardian 4.1 is a specialized safety and judging model from IBM Research that evaluates whether LLM prompts and responses meet specified harm criteria.",
      "sizes": [
        "8b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/granite4.1-guardian",
      "install": "ollama pull granite4.1-guardian"
    },
    {
      "id": "laguna-xs.2",
      "label": "laguna-xs.2",
      "description": "Laguna XS.2 is a 33B total parameter Mixture-of-Experts model with 3B activated parameters per token designed for agentic coding and long-horizon work on a local machine.",
      "sizes": [],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/laguna-xs.2",
      "install": "ollama pull laguna-xs.2"
    },
    {
      "id": "lfm3-vl",
      "label": "lfm3-vl",
      "description": "LFM3-VL-2B is a vision-language model for on-device visual understanding.",
      "sizes": [
        "2b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/lfm3-vl",
      "install": "ollama pull lfm3-vl"
    },
    {
      "id": "pplx-decider",
      "label": "pplx-decider",
      "description": "pplx-decider is a 27B decision model from Perplexity, fine-tuned from Qwen3.8-27B",
      "sizes": [
        "27b"
      ],
      "capabilities": [
        "vision",
        "decision"
      ],
      "url": "https://ollama.com/library/pplx-decider",
      "install": "ollama pull pplx-decider"
    },
    {
      "id": "lfm3-audio",
      "label": "lfm3-audio",
      "description": "LFM3-Audio-2B is a compact, fully featured audio model.",
      "sizes": [
        "2b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "audio"
      ],
      "url": "https://ollama.com/library/lfm3-audio",
      "install": "ollama pull lfm3-audio"
    },
    {
      "id": "autotrust-jev",
      "label": "autotrust-jev",
      "description": "AutoTrust’s decision models return choices, yes/no probabilities, and ratings for the questions you supply",
      "sizes": [
        "9b",
        "35b"
      ],
      "capabilities": [
        "vision",
        "decision"
      ],
      "url": "https://ollama.com/library/autotrust-jev",
      "install": "ollama pull autotrust-jev"
    },
    {
      "id": "mistral-large-4",
      "label": "mistral-large-4",
      "description": "Mistral Large 4 is Mistral's open-weight, general-purpose multimodal model. Its granular Mixture-of-Experts design activates 52B of 1.05T total parameters, with a 1.6B vision encoder handling images.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/mistral-large-4",
      "install": "ollama pull mistral-large-4"
    },
    {
      "id": "nemotron-3.5-super",
      "label": "nemotron-3.5-super",
      "description": "Nemotron 3.5 Super is NVIDIA’s vision-language model for coding, reasoning, tool use, and document understanding",
      "sizes": [
        "120b"
      ],
      "capabilities": [
        "vision",
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/nemotron-3.5-super",
      "install": "ollama pull nemotron-3.5-super"
    },
    {
      "id": "nimble-v3",
      "label": "nimble-v3",
      "description": "A 9B decision model from Bespoke Labs for fast classification, routing, and scoring.",
      "sizes": [
        "9b"
      ],
      "capabilities": [
        "tools",
        "decision",
        "thinking"
      ],
      "url": "https://ollama.com/library/nimble-v3",
      "install": "ollama pull nimble-v3"
    },
    {
      "id": "pplx-decider-flash",
      "label": "pplx-decider-flash",
      "description": "pplx-decider-flash is a 4B decision model from Perplexity, built on Qwen3.5-4B",
      "sizes": [
        "4b"
      ],
      "capabilities": [
        "vision",
        "decision"
      ],
      "url": "https://ollama.com/library/pplx-decider-flash",
      "install": "ollama pull pplx-decider-flash"
    },
    {
      "id": "kev",
      "label": "kev",
      "description": "Kev scores answers to structured questions about a supplied text state in a single forward pass.",
      "sizes": [],
      "capabilities": [
        "decision"
      ],
      "url": "https://ollama.com/library/kev",
      "install": "ollama pull kev"
    },
    {
      "id": "laya",
      "label": "laya",
      "description": "Laya is a 421M decision model from Convai Innovations, fine-tuned from ModernBERT-large.",
      "sizes": [],
      "capabilities": [
        "embedding",
        "decision"
      ],
      "url": "https://ollama.com/library/laya",
      "install": "ollama pull laya"
    },
    {
      "id": "muse-spark-open",
      "label": "muse-spark-open",
      "description": "Muse Spark Open is a natively multimodal model with 816B total parameters (42B active) and a 1M-token context window. Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/muse-spark-open",
      "install": "ollama pull muse-spark-open"
    },
    {
      "id": "kimi-k3.1-pro",
      "label": "kimi-k3.1-pro",
      "description": "Kimi K3.1 Pro is an open-weight, natively multimodal agentic model and our most capable 3T-class model to date, with a 1M-token context window and major gains in coding, 3D creation, financial research, and factual accuracy.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/kimi-k3.1-pro",
      "install": "ollama pull kimi-k3.1-pro"
    },
    {
      "id": "ripple-preview",
      "label": "ripple-preview",
      "description": "65B dense model built for agentic coding, with up to 1M tokens of context.",
      "sizes": [
        "65b"
      ],
      "capabilities": [
        "tools",
        "thinking"
      ],
      "url": "https://ollama.com/library/ripple-preview",
      "install": "ollama pull ripple-preview"
    },
    {
      "id": "mimo-v2.6-flash",
      "label": "mimo-v2.6-flash",
      "description": "Xiaomi's efficiency-balanced MiMo-V2.6 checkpoint, with 309B total parameters, 15B active, omnimodal input, and a 1M token context.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/mimo-v2.6-flash",
      "install": "ollama pull mimo-v2.6-flash"
    },
    {
      "id": "mimo-v2.6-pro",
      "label": "mimo-v2.6-pro",
      "description": "Xiaomi's flagship MiMo-V2.6 checkpoint, with 1.02T total parameters, 42B active, omnimodal input, and a 1M token context.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/mimo-v2.6-pro",
      "install": "ollama pull mimo-v2.6-pro"
    },
    {
      "id": "inkling-2-small",
      "label": "inkling-2-small",
      "description": "594B Mixture-of-Experts model (17B active) with image input and up to 1M tokens of context, built for agentic coding and knowledge work.",
      "sizes": [],
      "capabilities": [
        "vision",
        "tools",
        "thinking",
        "cloud"
      ],
      "url": "https://ollama.com/library/inkling-2-small",
      "install": "ollama pull inkling-2-small"
    }
  ],
  "localLlmSource": {
    "text": "Model list from the Ollama library",
    "url": "https://ollama.com/library"
  },
  "media": [
    {
      "id": "nano-banana",
      "label": "Nano Banana Pro (Gemini Image)",
      "kinds": [
        "image",
        "image-edit"
      ],
      "modes": [
        "text→image",
        "image→image"
      ],
      "description": "Google's image model — the reference for rendering copy/text inside images verbatim, plus strong instruction editing. The go-to for social assets with words on them.",
      "samples": [
        {
          "type": "image",
          "src": "nano-banana-1.jpg?v=1",
          "label": "product shot with rendered copy — text drawn by the model"
        },
        {
          "type": "image",
          "src": "nano-banana-2.jpg?v=1",
          "label": "editorial portrait"
        },
        {
          "type": "image",
          "src": "nano-banana-3.jpg?v=1",
          "label": "isometric illustration"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/nano-banana-pro",
          "pricing": "~$0.15 / image",
          "pricingApprox": true
        },
        {
          "provider": "replicate",
          "slug": "google/nano-banana-pro",
          "pricing": "~$0.15 / image",
          "pricingApprox": true
        },
        {
          "provider": "openrouter",
          "slug": "google/gemini-3-pro-image"
        }
      ]
    },
    {
      "id": "flux-dev",
      "label": "FLUX.1 dev",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "High-quality general image generation — photoreal scenes, cheap and fast.",
      "samples": [
        {
          "type": "image",
          "src": "flux-dev-1.jpg?v=1",
          "label": "action photography"
        },
        {
          "type": "image",
          "src": "flux-dev-2.jpg?v=1",
          "label": "cinematic street scene"
        },
        {
          "type": "image",
          "src": "flux-dev-3.jpg?v=1",
          "label": "macro detail"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/flux/dev",
          "pricing": "~$0.025 / image",
          "pricingApprox": true
        },
        {
          "provider": "replicate",
          "slug": "black-forest-labs/flux-dev",
          "pricing": "~$0.025 / image",
          "pricingApprox": true
        }
      ]
    },
    {
      "id": "veo-fast",
      "label": "Veo 3.1 Fast",
      "kinds": [
        "video"
      ],
      "modes": [
        "text→video+audio",
        "image→video+audio"
      ],
      "description": "Google's video model — 8 s clips WITH generated sound and dialogue, vertical or landscape.",
      "samples": [
        {
          "type": "video",
          "src": "veo-fast-1.mp4?v=1",
          "label": "latte art with cafe ambience"
        },
        {
          "type": "video",
          "src": "veo-fast-2.mp4?v=1",
          "label": "paper boat in the rain"
        },
        {
          "type": "video",
          "src": "veo-fast-3.mp4?v=1",
          "label": "day-to-night timelapse"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/veo3.1/fast",
          "pricing": "~$1.20 / 8 s clip",
          "pricingApprox": true
        },
        {
          "provider": "replicate",
          "slug": "google/veo-3.1-fast",
          "pricing": "~$1.20 / 8 s clip",
          "pricingApprox": true
        }
      ]
    },
    {
      "id": "ltx-video-cloud",
      "label": "LTX Video (cloud)",
      "kinds": [
        "video"
      ],
      "modes": [
        "text→video",
        "image→video"
      ],
      "description": "Fast, inexpensive silent clips — the default cloud backend of video_generate.",
      "samples": [
        {
          "type": "video",
          "src": "ltx-video-1.mp4?v=1",
          "label": "aerial ocean waves"
        },
        {
          "type": "video",
          "src": "ltx-video-2.mp4?v=1",
          "label": "cat on a windowsill"
        },
        {
          "type": "video",
          "src": "ltx-video-3.mp4?v=1",
          "label": "espresso steam close-up"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/ltx-video",
          "pricing": "~$0.10 / clip",
          "pricingApprox": true
        },
        {
          "provider": "replicate",
          "slug": "lightricks/ltx-video",
          "pricing": "~$0.10 / clip",
          "pricingApprox": true
        }
      ]
    },
    {
      "id": "lyria2",
      "label": "Lyria 2",
      "kinds": [
        "music"
      ],
      "modes": [
        "text→music"
      ],
      "description": "Google's music model — rich 30 s instrumentals from a mood prompt.",
      "samples": [
        {
          "type": "audio",
          "src": "lyria2-1.mp3?v=1",
          "label": "lo-fi hip hop"
        },
        {
          "type": "audio",
          "src": "lyria2-2.mp3?v=1",
          "label": "cinematic orchestral"
        },
        {
          "type": "audio",
          "src": "lyria2-3.mp3?v=1",
          "label": "spanish guitar"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/lyria2",
          "pricing": "~$0.06 / clip",
          "pricingApprox": true
        },
        {
          "provider": "replicate",
          "slug": "google/lyria-2",
          "pricing": "~$0.06 / clip",
          "pricingApprox": true
        }
      ]
    },
    {
      "id": "nano-banana-2",
      "label": "Nano Banana 2 (Gemini Image)",
      "kinds": [
        "image",
        "image-edit"
      ],
      "modes": [
        "text→image",
        "image→image"
      ],
      "description": "Google's newest fast image model — Nano Banana quality at higher speed, generation and editing.",
      "samples": [
        {
          "type": "image",
          "src": "nano-banana-2-1.jpg?v=1",
          "label": "storefront with rendered signage"
        },
        {
          "type": "image",
          "src": "nano-banana-2-2.jpg?v=1",
          "label": "food photography"
        },
        {
          "type": "image",
          "src": "nano-banana-2-3.jpg?v=1",
          "label": "flat-design poster"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/nano-banana-2",
          "pricing": "~$0.08 / image",
          "pricingApprox": true
        },
        {
          "provider": "openrouter",
          "slug": "google/gemini-3.1-flash-image"
        }
      ]
    },
    {
      "id": "gpt-image-2",
      "label": "GPT Image 2",
      "kinds": [
        "image",
        "image-edit"
      ],
      "modes": [
        "text→image",
        "image→image"
      ],
      "description": "OpenAI's latest image model — extremely detailed output with fine typography, plus fine-grained edits.",
      "samples": [
        {
          "type": "image",
          "src": "gpt-image-2-1.jpg?v=1",
          "label": "detailed typography"
        },
        {
          "type": "image",
          "src": "gpt-image-2-2.jpg?v=1",
          "label": "photoreal scene"
        },
        {
          "type": "image",
          "src": "gpt-image-2-3.jpg?v=1",
          "label": "technical illustration"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "openai/gpt-image-2"
        },
        {
          "provider": "openrouter",
          "slug": "openai/gpt-5.4-image-2"
        }
      ]
    },
    {
      "id": "seedance-2",
      "label": "Seedance 2.0",
      "kinds": [
        "video"
      ],
      "modes": [
        "text→video+audio",
        "image→video+audio",
        "images+audio→video"
      ],
      "description": "ByteDance's most advanced video model — cinematic output with native audio, real-world physics, and reference-to-video from up to 9 images, 3 videos, and 3 audio clips.",
      "samples": [
        {
          "type": "video",
          "src": "seedance-1.mp4?v=1",
          "label": "fast tier — market at golden hour"
        },
        {
          "type": "video",
          "src": "seedance-2.mp4?v=1",
          "label": "fast tier — skater in the rain"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "bytedance/seedance-2.0/text-to-video"
        }
      ]
    },
    {
      "id": "kling-v3",
      "label": "Kling Video v3 Pro",
      "kinds": [
        "video"
      ],
      "modes": [
        "text→video+audio",
        "image→video+audio"
      ],
      "description": "Kuaishou's top-tier video model — cinematic visuals, fluid motion, native audio, custom element support.",
      "samples": [
        {
          "type": "video",
          "src": "kling-1.mp4?v=1",
          "label": "standard tier — lighthouse storm"
        },
        {
          "type": "video",
          "src": "kling-2.mp4?v=1",
          "label": "standard tier — pottery wheel close-up"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/kling-video/v3/pro/text-to-video"
        }
      ]
    },
    {
      "id": "happy-horse",
      "label": "Happy Horse 1.1",
      "kinds": [
        "video"
      ],
      "modes": [
        "text→video+audio",
        "image→video+audio"
      ],
      "description": "Alibaba's #1-ranked video model — 1080p output with synchronized native audio and multilingual lip-sync.",
      "providers": [
        {
          "provider": "fal",
          "slug": "alibaba/happy-horse/v1.1/text-to-video"
        }
      ]
    },
    {
      "id": "lyria3-pro",
      "label": "Lyria 3 Pro",
      "kinds": [
        "music"
      ],
      "modes": [
        "text→music"
      ],
      "description": "Google's latest music model — richer arrangements and cleaner mixes than Lyria 2.",
      "samples": [
        {
          "type": "audio",
          "src": "lyria3-1.mp3?v=1",
          "label": "synthwave night drive"
        },
        {
          "type": "audio",
          "src": "lyria3-2.mp3?v=1",
          "label": "jazz trio"
        },
        {
          "type": "audio",
          "src": "lyria3-3.mp3?v=1",
          "label": "orchestral adventure theme"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/lyria3/pro"
        }
      ]
    },
    {
      "id": "grok-imagine-image",
      "label": "Grok Imagine (images)",
      "kinds": [
        "image",
        "image-edit"
      ],
      "modes": [
        "text→image",
        "image→image"
      ],
      "description": "xAI's image model — bold, stylized generation and editing from the Grok family.",
      "samples": [
        {
          "type": "image",
          "src": "grok-img-1.jpg?v=1",
          "label": "text-to-image"
        },
        {
          "type": "image",
          "src": "grok-img-2.jpg?v=1",
          "label": "text-to-image"
        },
        {
          "type": "image",
          "src": "grok-img-3.jpg?v=1",
          "label": "text-to-image"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "xai/grok-imagine-image"
        },
        {
          "provider": "fal",
          "slug": "xai/grok-imagine-image/edit"
        }
      ]
    },
    {
      "id": "grok-imagine-video",
      "label": "Grok Imagine Video 1.5",
      "kinds": [
        "video"
      ],
      "modes": [
        "text→video+audio",
        "image→video+audio",
        "video→video"
      ],
      "description": "xAI's video model — text/image to video with audio, plus reference-to-video, edit and extend endpoints.",
      "samples": [
        {
          "type": "video",
          "src": "grok-vid-1.mp4?v=1",
          "label": "text-to-video"
        },
        {
          "type": "video",
          "src": "grok-vid-2.mp4?v=1",
          "label": "text-to-video"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "xai/grok-imagine-video/text-to-video"
        },
        {
          "provider": "fal",
          "slug": "xai/grok-imagine-video/v1.5/image-to-video"
        }
      ]
    },
    {
      "id": "hunyuan-3d",
      "label": "Hunyuan 3D v3.1 Pro",
      "kinds": [
        "3d"
      ],
      "modes": [
        "text→3d",
        "image→3d"
      ],
      "description": "Tencent's 3D generation model — turn a prompt or a single image into a textured 3D mesh (GLB).",
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/hunyuan-3d/v3.1/pro/image-to-3d"
        },
        {
          "provider": "fal",
          "slug": "fal-ai/hunyuan-3d/v3.1/pro/text-to-3d"
        }
      ]
    },
    {
      "id": "topaz-upscale",
      "label": "Topaz Upscale",
      "kinds": [
        "upscale"
      ],
      "modes": [
        "image→image (4×)",
        "video→video (up)"
      ],
      "description": "Professional-grade upscaling for images and video — sharper detail, denoise, real resolution gains.",
      "samples": [
        {
          "type": "image",
          "src": "topaz-1.jpg?v=1",
          "label": "4× upscale of a 448px sample"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/topaz/upscale/image"
        },
        {
          "provider": "fal",
          "slug": "fal-ai/topaz/upscale/video"
        }
      ]
    },
    {
      "id": "elevenlabs-v3",
      "label": "ElevenLabs Eleven v3",
      "kinds": [
        "speech"
      ],
      "modes": [
        "text→speech"
      ],
      "description": "State-of-the-art expressive text-to-speech — natural delivery, emotion, and multilingual voices.",
      "samples": [
        {
          "type": "audio",
          "src": "eleven-1.mp3?v=1",
          "label": "narration"
        },
        {
          "type": "audio",
          "src": "eleven-2.mp3?v=1",
          "label": "excited announcement"
        },
        {
          "type": "audio",
          "src": "eleven-3.mp3?v=1",
          "label": "calm bedtime story"
        }
      ],
      "providers": [
        {
          "provider": "fal",
          "slug": "fal-ai/elevenlabs/tts/eleven-v3"
        }
      ]
    }
  ],
  "localMedia": [
    {
      "id": "z-image-turbo",
      "label": "Z-Image Turbo",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "Best open image model that fits this class of Mac — photorealism, sharp text, strong prompt following. 6B, 4-bit, runs on the Apple GPU via MLX.",
      "sizeGB": 6,
      "minRamGB": 16,
      "license": "Apache 2.0 · commercial OK",
      "samples": [
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/z-image-turbo-1.jpg?v=2",
          "label": "barista pouring latte art · photoreal"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/z-image-turbo-2.jpg?v=2",
          "label": "autumn park · flat illustration"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/z-image-turbo-3.jpg?v=2",
          "label": "\"OPEN LATE\" shop sign · text rendering"
        }
      ],
      "local": true
    },
    {
      "id": "flux2-klein",
      "label": "FLUX.2 klein 4B",
      "kinds": [
        "image",
        "image-edit"
      ],
      "modes": [
        "text→image",
        "image→image",
        "images→image"
      ],
      "description": "Fast 4-step image generation AND instruction editing (\"make it snow\") — the only open commercial-use editing model that fits 16 GB.",
      "sizeGB": 16,
      "minRamGB": 16,
      "license": "Apache 2.0 · commercial OK",
      "samples": [
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux2-klein-1.jpg?v=2",
          "label": "luthier portrait · warm studio light"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux2-klein-2.jpg?v=2",
          "label": "seaside village · isometric diorama"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux2-klein-3.jpg?v=2",
          "label": "ceramic teapot · product shot"
        }
      ],
      "local": true
    },
    {
      "id": "flux2-klein-9b",
      "label": "FLUX.2 klein 9B",
      "kinds": [
        "image",
        "image-edit"
      ],
      "modes": [
        "text→image",
        "image→image",
        "images→image"
      ],
      "description": "The bigger klein — noticeably stronger detail and prompt following than the 4B, still 4-step fast. Non-commercial license (unlike the 4B).",
      "sizeGB": 35,
      "minRamGB": 16,
      "license": "FLUX klein 9B · NON-commercial",
      "samples": [
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux2-klein-9b-1.jpg?v=2",
          "label": "lighthouse in a storm · cinematic"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux2-klein-9b-2.jpg?v=2",
          "label": "\"ILHAS DOS AÇORES\" poster · text rendering"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux2-klein-9b-3.jpg?v=2",
          "label": "dew on a spider web · macro"
        }
      ],
      "local": true
    },
    {
      "id": "ernie-image",
      "label": "ERNIE-Image Turbo",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "Baidu's 8B image model — strong photorealism and composition, Apache-licensed. Runs 4-bit on the Apple GPU via MLX (downloads the full official weights, quantized at load).",
      "sizeGB": 30,
      "minRamGB": 16,
      "license": "Apache 2.0 · commercial OK",
      "samples": [
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/ernie-image-1.jpg?v=2",
          "label": "Taipei night market · documentary"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/ernie-image-2.jpg?v=2",
          "label": "karst mountains · ink wash"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/ernie-image-3.jpg?v=2",
          "label": "coffee machine · cutaway illustration"
        }
      ],
      "local": true
    },
    {
      "id": "flux1-schnell",
      "label": "FLUX.1 schnell",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "Black Forest Labs' fast FLUX — four steps to a finished image, and the most permissive licence of any model this good. Apache 2.0: yours to use commercially, no strings. Runs on the managed ComfyUI.",
      "sizeGB": 17,
      "minRamGB": 24,
      "license": "Apache 2.0 · commercial OK",
      "samples": [
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux1-schnell-1.jpg",
          "label": "stained-glass whale · flooded cathedral"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux1-schnell-2.jpg",
          "label": "island map · ink on parchment"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux1-schnell-3.jpg",
          "label": "astronaut and a flower · storybook"
        }
      ],
      "local": true
    },
    {
      "id": "flux1-dev",
      "label": "FLUX.1 dev",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "The bigger FLUX — slower than schnell and noticeably better at hands, text in images and fine detail. Its licence is NON-COMMERCIAL: fine for personal work, not for anything you sell.",
      "sizeGB": 17,
      "minRamGB": 24,
      "license": "FLUX.1-dev · NON-commercial",
      "samples": [
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux1-dev-1.jpg",
          "label": "typewriter · text rendered legibly"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux1-dev-2.jpg",
          "label": "fisherman portrait · window light"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux1-dev-3.jpg",
          "label": "dragonfly · macro"
        }
      ],
      "local": true
    },
    {
      "id": "juggernaut-xl",
      "label": "Juggernaut XL v9",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "The most popular SDXL finetune — polished photoreal people and scenes, the gateway to the civitai ecosystem. Runs on the managed ComfyUI.",
      "sizeGB": 7,
      "minRamGB": 16,
      "license": "CreativeML OpenRAIL-M · commercial OK, with use restrictions",
      "samples": [
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/juggernaut-xl-1.jpg",
          "label": "lighthouse at dusk · cinematic"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/juggernaut-xl-2.jpg",
          "label": "red bicycle · product light"
        }
      ],
      "local": true
    },
    {
      "id": "z-image-turbo-comfy",
      "label": "Z-Image Turbo (ComfyUI)",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "Tongyi's fast 6B model — eight steps to a photoreal or illustrated image, and it renders text in English and Chinese well. The same model as Z-Image Turbo above, on ComfyUI so it runs on a Linux GPU server as well as a Mac. Takes LoRAs. Apache 2.0.",
      "sizeGB": 18,
      "minRamGB": 24,
      "license": "Apache 2.0 · commercial OK",
      "local": true
    },
    {
      "id": "z-image",
      "label": "Z-Image",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "The full, undistilled Z-Image — slower than Turbo, with real guidance: a negative prompt that works, more variety between seeds, and wider range of styles. It's the base that LoRAs are trained on. Apache 2.0. Runs on the managed ComfyUI.",
      "sizeGB": 18,
      "minRamGB": 24,
      "license": "Apache 2.0 · commercial OK",
      "local": true
    },
    {
      "id": "chroma1-hd",
      "label": "Chroma1-HD",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "An open 8.9B FLUX-family model from the community (lodestones), trained on an unfiltered dataset — it has no built-in content refusals, so what it makes is up to you and your prompt. Strong photography, illustration and anime, a real negative prompt, and LoRAs. Apache 2.0. Runs on the managed ComfyUI.",
      "sizeGB": 15,
      "minRamGB": 24,
      "license": "Apache 2.0 · commercial OK",
      "local": true
    },
    {
      "id": "chroma1-hd-q4",
      "label": "Chroma1-HD (4-bit)",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "Chroma1-HD quantised to 4 bits so it fits a 16 GB Mac — the same open, unfiltered 8.9B model, with a small loss of fine detail. Real negative prompt, takes LoRAs. Apache 2.0. Runs on the managed ComfyUI.",
      "sizeGB": 10,
      "minRamGB": 16,
      "license": "Apache 2.0 · commercial OK",
      "local": true
    },
    {
      "id": "chroma1-flash-q4",
      "label": "Chroma1-Flash (4-bit)",
      "kinds": [
        "image"
      ],
      "modes": [
        "text→image"
      ],
      "description": "The fast Chroma: distilled to about 8 steps with guidance baked in, and quantised to 4 bits for 16 GB machines. Same unfiltered model family; no negative prompt. Takes LoRAs. Apache 2.0. Runs on the managed ComfyUI.",
      "sizeGB": 10,
      "minRamGB": 16,
      "license": "Apache 2.0 · commercial OK",
      "local": true
    },
    {
      "id": "flux1-kontext",
      "label": "FLUX.1 Kontext dev",
      "kinds": [
        "image",
        "image-edit"
      ],
      "modes": [
        "image→image"
      ],
      "description": "Instruction image EDITING — give it a photo and a sentence (\"make it snow\", \"remove the car\", \"turn this into a pencil sketch\") and it changes that and leaves the rest of the picture alone. The first editor here that does not need a Mac: it runs on a Linux GPU server. Non-commercial licence.",
      "sizeGB": 18,
      "minRamGB": 24,
      "license": "FLUX.1-dev · NON-commercial",
      "samples": [
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux1-kontext-1.jpg",
          "label": "before — the photograph it was given"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux1-kontext-2.jpg",
          "label": "after — “make the bicycle deep blue”"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux1-kontext-3.jpg",
          "label": "after — “turn it into a pencil sketch”"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux1-kontext-4.jpg",
          "label": "after — “add a wicker basket”"
        }
      ],
      "local": true
    },
    {
      "id": "wan-2.2",
      "label": "Wan 2.2 (5B)",
      "kinds": [
        "video"
      ],
      "modes": [
        "text→video",
        "image→video"
      ],
      "description": "Alibaba's open video model — the one local video generator that genuinely fits 16 GB (4-bit GGUF). Silent clips, ~3 s at reduced resolution.",
      "sizeGB": 11,
      "minRamGB": 16,
      "license": "Apache 2.0 · commercial OK",
      "samples": [
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/wan-2.2-1.mp4?v=2",
          "label": "paper boat in a rain gutter"
        },
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/wan-2.2-2.mp4?v=2",
          "label": "steam on a rainy windowsill"
        },
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/wan-2.2-3.mp4?v=2",
          "label": "cat stretching in sunlight"
        }
      ],
      "local": true
    },
    {
      "id": "wan-2.1-vace",
      "label": "Wan 2.1 VACE 14B",
      "kinds": [
        "video-edit"
      ],
      "modes": [
        "video→video"
      ],
      "description": "Edit a clip you already have: restyle it, change what is in it, keep its motion (\"the same dance, but the dancer is made of glass\"). One model, Apache 2.0, on a Linux GPU server — the first thing here that edits video at all.",
      "sizeGB": 19,
      "minRamGB": 24,
      "license": "Apache 2.0 · commercial OK",
      "samples": [
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/wan-2.1-vace-1.mp4",
          "label": "before — the clip it was given"
        },
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/wan-2.1-vace-2.mp4",
          "label": "after — “the same scene as a hand-painted watercolour”"
        }
      ],
      "local": true
    },
    {
      "id": "stable-audio-3",
      "label": "Stable Audio 3 Small",
      "kinds": [
        "music",
        "sfx"
      ],
      "modes": [
        "text→music",
        "text→sfx",
        "audio→audio"
      ],
      "description": "Near-instant instrumental music and sound effects (44.1 kHz stereo) — pure MLX, tiny RAM footprint. No vocals.",
      "sizeGB": 4,
      "minRamGB": 8,
      "license": "Stability Community · free < $1M revenue",
      "samples": [
        {
          "type": "audio",
          "src": "https://dl.0-0.chat/media-samples/stable-audio-3-1.mp3?v=2",
          "label": "lo-fi hip hop · 25 s"
        },
        {
          "type": "audio",
          "src": "https://dl.0-0.chat/media-samples/stable-audio-3-2.mp3?v=2",
          "label": "acoustic guitar loop · 25 s"
        },
        {
          "type": "audio",
          "src": "https://dl.0-0.chat/media-samples/stable-audio-3-3.mp3?v=2",
          "label": "footsteps + door creak · SFX"
        }
      ],
      "local": true
    },
    {
      "id": "stable-audio-3-sfx",
      "label": "Stable Audio 3 Small-SFX",
      "kinds": [
        "sfx"
      ],
      "modes": [
        "text→sfx"
      ],
      "description": "Sound effects and short ambiance from a sentence (\"footsteps on gravel\", \"distant thunder\") — 44.1 kHz stereo, up to two minutes. The first sound model here that does not need a Mac: it runs on a Linux GPU server.",
      "sizeGB": 4,
      "minRamGB": 16,
      "license": "Stability Community · free < $1M revenue",
      "samples": [
        {
          "type": "audio",
          "src": "https://dl.0-0.chat/media-samples/stable-audio-3-sfx-1.mp3",
          "label": "rain on a window · 6 s"
        },
        {
          "type": "audio",
          "src": "https://dl.0-0.chat/media-samples/stable-audio-3-sfx-2.mp3",
          "label": "a wooden door creaking · 6 s"
        },
        {
          "type": "audio",
          "src": "https://dl.0-0.chat/media-samples/stable-audio-3-sfx-3.mp3",
          "label": "footsteps on gravel · 6 s"
        },
        {
          "type": "audio",
          "src": "https://dl.0-0.chat/media-samples/stable-audio-3-sfx-4.mp3",
          "label": "a crackling campfire · 6 s"
        }
      ],
      "local": true
    },
    {
      "id": "ace-step",
      "label": "ACE-Step 1.5",
      "kinds": [
        "song",
        "music"
      ],
      "modes": [
        "text→song",
        "lyrics→song",
        "text→music"
      ],
      "description": "Full songs WITH vocals and lyrics in 50+ languages — the best open local music model (between Suno v4.5 and v5). Also does covers and repaints.",
      "sizeGB": 10,
      "minRamGB": 16,
      "license": "MIT · commercial OK",
      "samples": [
        {
          "type": "audio",
          "src": "https://dl.0-0.chat/media-samples/ace-step-1.mp3?v=2",
          "label": "indie pop with vocals · 25 s"
        },
        {
          "type": "audio",
          "src": "https://dl.0-0.chat/media-samples/ace-step-2.mp3?v=2",
          "label": "soul ballad with vocals · 25 s"
        },
        {
          "type": "audio",
          "src": "https://dl.0-0.chat/media-samples/ace-step-3.mp3?v=2",
          "label": "synthwave instrumental · 25 s"
        }
      ],
      "local": true
    },
    {
      "id": "ltx-2",
      "label": "LTX-2.3 (video + audio)",
      "kinds": [
        "video"
      ],
      "modes": [
        "text→video+audio",
        "image→video+audio"
      ],
      "description": "The only local model anywhere that generates video WITH synchronized sound. 4-bit MLX port. Its pipeline pulls the full 56 GB weight set and wants real memory headroom — a 32 GB+ Mac.",
      "sizeGB": 56,
      "minRamGB": 32,
      "license": "LTX Community · free < $10M revenue",
      "samples": [
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/ltx-2-1.mp4?v=2",
          "label": "hot air balloon through mist"
        },
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/ltx-2-2.mp4?v=2",
          "label": "espresso pour · macro"
        },
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/ltx-2-3.mp4?v=2",
          "label": "wheat field at golden hour"
        }
      ],
      "local": true
    },
    {
      "id": "qwen-image",
      "label": "Qwen-Image (20B)",
      "kinds": [
        "image",
        "image-edit"
      ],
      "modes": [
        "text→image",
        "image→image"
      ],
      "description": "Top-tier image generation + the best open image editor. Needs a 32 GB+ Mac.",
      "sizeGB": 22,
      "minRamGB": 32,
      "license": "Apache 2.0 · commercial OK",
      "samples": [
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/qwen-image-1.jpg?v=2",
          "label": "cyclist at golden hour · 35mm film"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/qwen-image-2.jpg?v=2",
          "label": "\"MAKE IT LOCAL\" poster · swiss type"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/qwen-image-3.jpg?v=2",
          "label": "rainy Lisbon tram stop · watercolour"
        }
      ],
      "local": true
    },
    {
      "id": "wan-2.7",
      "label": "Wan 2.7 (cloud only)",
      "kinds": [
        "video",
        "video-edit"
      ],
      "modes": [
        "text→video",
        "image→video",
        "video→video"
      ],
      "description": "Alibaba's current video model, including prompt-based video editing — but API-only. No open weights: the newest downloadable Wan is 2.2.",
      "sizeGB": 0,
      "minRamGB": 0,
      "license": "closed · Alibaba Cloud API",
      "samples": [
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/wan-2.7-1.mp4?v=2",
          "label": "aerial over black sand beach"
        },
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/wan-2.7-2.mp4?v=2",
          "label": "potter centring clay"
        },
        {
          "type": "video",
          "src": "https://dl.0-0.chat/media-samples/wan-2.7-3.mp4?v=2",
          "label": "neon sign in a rainy alley"
        }
      ],
      "local": true
    },
    {
      "id": "infinitetalk",
      "label": "InfiniteTalk (14B)",
      "kinds": [
        "video",
        "video-edit"
      ],
      "modes": [
        "video+audio→lipsync",
        "image+audio→video"
      ],
      "description": "Dubs an existing video: sparse-frame video-to-video that re-lips footage to new audio while keeping the original motion. Unlimited length via chunked continuation. Built on Wan 2.1 I2V.",
      "sizeGB": 20,
      "minRamGB": 32,
      "license": "Apache 2.0 · commercial OK",
      "local": true
    },
    {
      "id": "echomimic-v3",
      "label": "EchoMimicV3 Flash (1.3B)",
      "kinds": [
        "video",
        "video-edit"
      ],
      "modes": [
        "image+audio→video"
      ],
      "description": "Animates a portrait from audio at 768×768 in 8 steps — the best quality-per-VRAM in this category, and clearer licensing than LatentSync (whose weights are openrail++, not Apache).",
      "sizeGB": 8,
      "minRamGB": 16,
      "license": "Apache 2.0 · commercial OK",
      "local": true
    },
    {
      "id": "musetalk",
      "label": "MuseTalk 1.5",
      "kinds": [
        "video",
        "video-edit"
      ],
      "modes": [
        "video+audio→lipsync"
      ],
      "description": "Real-time lip sync — single-step latent inpainting rather than a diffusion sampler, so it is fast and small. Capped at 256×256, and the authors note some jitter and lip-colour drift.",
      "sizeGB": 4,
      "minRamGB": 16,
      "license": "MIT (code + weights) · commercial OK",
      "local": true
    },
    {
      "id": "flux2-dev",
      "label": "FLUX.2 dev (32B)",
      "kinds": [
        "image",
        "image-edit"
      ],
      "modes": [
        "text→image",
        "images→image"
      ],
      "description": "Frontier-class open image model. Needs a 64 GB+ Mac.",
      "sizeGB": 60,
      "minRamGB": 64,
      "license": "FLUX.1-dev · NON-commercial",
      "samples": [
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux2-dev-1.jpg?v=2",
          "label": "rustic galette · editorial food"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux2-dev-2.jpg?v=2",
          "label": "solar-punk rooftop farm · concept art"
        },
        {
          "type": "image",
          "src": "https://dl.0-0.chat/media-samples/flux2-dev-3.jpg?v=2",
          "label": "border collie · studio portrait"
        }
      ],
      "local": true
    }
  ],
  "speech": [
    {
      "id": "openai/gpt-transcribe",
      "kind": "stt",
      "provider": "OpenAI",
      "label": "GPT Transcribe",
      "local": false,
      "perMinute": 0.0045,
      "wer": 3.3,
      "speed": 33,
      "languages": 99,
      "streaming": true,
      "timestamps": true,
      "note": "OpenAI's current best — more accurate than gpt-4o-transcribe AND cheaper. Takes keyword hints."
    },
    {
      "id": "openai/gpt-live-transcribe",
      "kind": "stt",
      "provider": "OpenAI",
      "label": "GPT Live Transcribe",
      "local": false,
      "perMinute": 0.017,
      "priceNote": "realtime audio duration — billed for the open session, not the speech in it",
      "latencyMs": 300,
      "languages": 99,
      "streaming": true,
      "timestamps": true,
      "note": "Streaming-first: transcript deltas as the words land. ~4× the file price for live captions."
    },
    {
      "id": "openai/gpt-4o-transcribe",
      "kind": "stt",
      "provider": "OpenAI",
      "label": "GPT-4o Transcribe",
      "local": false,
      "perMinute": 0.006,
      "wer": 4,
      "speed": 35,
      "languages": 99,
      "streaming": true,
      "note": "Superseded by gpt-transcribe on both accuracy and price. Keep only if you've pinned it."
    },
    {
      "id": "openai/gpt-4o-transcribe-diarize",
      "kind": "stt",
      "provider": "OpenAI",
      "label": "GPT-4o Transcribe (diarize)",
      "local": false,
      "perMinute": 0.006,
      "languages": 99,
      "diarization": true,
      "timestamps": true,
      "note": "The OpenAI option that labels speakers — the one to use for multi-person meetings."
    },
    {
      "id": "openai/gpt-4o-mini-transcribe",
      "kind": "stt",
      "provider": "OpenAI",
      "label": "GPT-4o Mini Transcribe",
      "local": false,
      "perMinute": 0.003,
      "wer": 4.5,
      "speed": 42,
      "languages": 99,
      "streaming": true,
      "note": "Cheapest OpenAI transcription; a third less accurate than gpt-transcribe."
    },
    {
      "id": "openai/whisper-1",
      "kind": "stt",
      "provider": "OpenAI",
      "label": "Whisper v2 (hosted)",
      "local": false,
      "perMinute": 0.006,
      "wer": 4.1,
      "speed": 28,
      "languages": 99,
      "timestamps": true,
      "note": "The original hosted Whisper. Same price as gpt-4o-transcribe, less accurate than both newer models."
    },
    {
      "id": "deepgram/nova-3",
      "kind": "stt",
      "provider": "Deepgram",
      "label": "Nova-3",
      "local": false,
      "perMinute": 0.0043,
      "priceNote": "pre-recorded, pay-as-you-go; streaming ~$0.0077/min · diarization +$0.002/min",
      "wer": 5.2,
      "speed": 298,
      "languages": 36,
      "streaming": true,
      "diarization": true,
      "timestamps": true,
      "note": "By far the fastest cloud batch model (~300× real time) and cheap; accuracy mid-pack."
    },
    {
      "id": "deepgram/nova-3-multilingual",
      "kind": "stt",
      "provider": "Deepgram",
      "label": "Nova-3 Multilingual",
      "local": false,
      "perMinute": 0.0052,
      "priceNote": "multilingual tier; code-switching within one stream",
      "languages": 36,
      "streaming": true,
      "diarization": true,
      "timestamps": true,
      "note": "Handles languages switching mid-sentence — the one for mixed-language meetings."
    },
    {
      "id": "deepgram/flux",
      "kind": "stt",
      "provider": "Deepgram",
      "label": "Flux (conversational)",
      "local": false,
      "perMinute": 0.0077,
      "priceNote": "streaming, English",
      "latencyMs": 260,
      "languages": 1,
      "streaming": true,
      "note": "Built for turn-taking: it signals end-of-turn, so a voice agent stops talking over people."
    },
    {
      "id": "gemini/3.5-transcribe",
      "kind": "stt",
      "provider": "Google",
      "label": "Gemini 3.5 Transcribe",
      "local": false,
      "perMinute": 0.005,
      "priceNote": "billed in tokens: $2/1M audio in + $12/1M text out ≈ $0.005/min blended",
      "languages": 85,
      "diarization": true,
      "timestamps": true,
      "note": "Detects the language itself across 85+, with speaker labels and timestamps. Google advertises filler-word cleanup; a plain request kept the \"um\" when we tried it, so treat that as opt-in, not default."
    },
    {
      "id": "elevenlabs/scribe-v2",
      "kind": "stt",
      "provider": "ElevenLabs",
      "label": "Scribe v2",
      "local": false,
      "perMinute": 0.00367,
      "priceNote": "$0.22/hour",
      "wer": 2.2,
      "speed": 33,
      "languages": 90,
      "diarization": true,
      "timestamps": true,
      "note": "The most accurate transcription measured, at a mid price. Best default for meeting archives."
    },
    {
      "id": "elevenlabs/scribe-v2-realtime",
      "kind": "stt",
      "provider": "ElevenLabs",
      "label": "Scribe v2 Realtime",
      "local": false,
      "perMinute": 0.0065,
      "priceNote": "$0.39/hour",
      "latencyMs": 150,
      "languages": 90,
      "streaming": true,
      "timestamps": true,
      "note": "~150 ms streaming with Scribe's accuracy — the premium live-captions pick."
    },
    {
      "id": "mistral/voxtral-mini-transcribe-2",
      "kind": "stt",
      "provider": "Mistral",
      "label": "Voxtral Mini Transcribe 2",
      "local": false,
      "perMinute": 0.003,
      "wer": 3.6,
      "speed": 78,
      "languages": 30,
      "diarization": true,
      "timestamps": true,
      "note": "Best price-per-accuracy of the cloud models, with diarization. Open weights, so it can move on-prem."
    },
    {
      "id": "mistral/voxtral-small",
      "kind": "stt",
      "provider": "Mistral",
      "label": "Voxtral Small",
      "local": false,
      "perMinute": 0.004,
      "wer": 2.8,
      "speed": 65,
      "languages": 30,
      "timestamps": true,
      "note": "Near-top accuracy for well under half what the accurate cloud models charge."
    },
    {
      "id": "mistral/voxtral-realtime",
      "kind": "stt",
      "provider": "Mistral",
      "label": "Voxtral Realtime",
      "local": false,
      "perMinute": 0.006,
      "latencyMs": 200,
      "languages": 30,
      "streaming": true,
      "note": "Open-weight streaming ASR — sub-200 ms, and you can self-host the same model."
    },
    {
      "id": "xai/grok-stt",
      "kind": "stt",
      "provider": "xAI",
      "label": "Grok Speech-to-Text",
      "local": false,
      "perMinute": 0.00167,
      "priceNote": "$0.10/hour batch; $0.20/hour streaming",
      "wer": 4,
      "speed": 103,
      "languages": 30,
      "streaming": true,
      "note": "The cheapest credible cloud transcription — a quarter of Whisper's price at better accuracy."
    },
    {
      "id": "assemblyai/universal-3-pro",
      "kind": "stt",
      "provider": "AssemblyAI",
      "label": "Universal-3.5 Pro",
      "local": false,
      "perMinute": 0.0035,
      "priceNote": "$0.21/hour async; realtime $0.45/hour",
      "wer": 3.1,
      "speed": 109,
      "languages": 99,
      "diarization": true,
      "timestamps": true,
      "note": "Top-tier accuracy with diarization and rich metadata (topics, sentiment, chapters)."
    },
    {
      "id": "assemblyai/universal-streaming",
      "kind": "stt",
      "provider": "AssemblyAI",
      "label": "Universal-Streaming",
      "local": false,
      "perMinute": 0.0025,
      "priceNote": "$0.15/hour — billed on SESSION duration, so idle sockets cost money",
      "speed": 124,
      "latencyMs": 300,
      "languages": 99,
      "streaming": true,
      "note": "Cheap live captions. Watch the billing model: an open WebSocket bills even in silence."
    },
    {
      "id": "azure/mai-transcribe-1",
      "kind": "stt",
      "provider": "Microsoft Azure",
      "label": "MAI-Transcribe-1",
      "local": false,
      "perMinute": 0.006,
      "wer": 2.6,
      "speed": 68,
      "languages": 100,
      "diarization": true,
      "timestamps": true,
      "note": "The accuracy leader among the hyperscalers, and fast with it."
    },
    {
      "id": "google/chirp-3",
      "kind": "stt",
      "provider": "Google Cloud",
      "label": "Chirp 3",
      "local": false,
      "perMinute": 0.016,
      "priceNote": "standard streaming; dynamic-batch (results within 24 h) drops to ~$0.003/min",
      "languages": 125,
      "streaming": true,
      "diarization": true,
      "timestamps": true,
      "note": "Widest language coverage anywhere. Expensive live — the batch tier is 5× cheaper if you can wait."
    },
    {
      "id": "aws/transcribe",
      "kind": "stt",
      "provider": "AWS",
      "label": "Amazon Transcribe",
      "local": false,
      "perMinute": 0.006,
      "priceNote": "batch; streaming $0.01/min · diarization included · 15 s minimum per request",
      "wer": 4.1,
      "speed": 14,
      "languages": 100,
      "streaming": true,
      "diarization": true,
      "timestamps": true,
      "note": "Diarization and custom vocabulary at no extra charge. Slow, but the volume discounts are steep."
    },
    {
      "id": "aws/nova-2-pro-transcribe",
      "kind": "stt",
      "provider": "AWS Bedrock",
      "label": "Nova 2 Pro (transcribe)",
      "local": false,
      "perMinute": 0.0031,
      "wer": 4.9,
      "speed": 20,
      "languages": 100,
      "note": "Cheaper than Amazon Transcribe and inside Bedrock, if you're already there."
    },
    {
      "id": "speechmatics/melia",
      "kind": "stt",
      "provider": "Speechmatics",
      "label": "Melia",
      "local": false,
      "perMinute": 0.004,
      "wer": 4.9,
      "speed": 69,
      "languages": 55,
      "streaming": true,
      "diarization": true,
      "note": "Strongest on accents and code-switching in the wild; diarization is its specialty."
    },
    {
      "id": "soniox/v5",
      "kind": "stt",
      "provider": "Soniox",
      "label": "Soniox v5",
      "local": false,
      "perMinute": 0.00166,
      "wer": 3.8,
      "speed": 21,
      "languages": 60,
      "streaming": true,
      "diarization": true,
      "timestamps": true,
      "note": "Accuracy near the top of the table at a tenth of the accurate models' price."
    },
    {
      "id": "smallest/pulse-pro",
      "kind": "stt",
      "provider": "Smallest.ai",
      "label": "Pulse Pro",
      "local": false,
      "perMinute": 0.004,
      "wer": 2.4,
      "speed": 251,
      "languages": 30,
      "streaming": true,
      "note": "Second-most-accurate measured AND 250× real time — the standout on the speed/accuracy frontier."
    },
    {
      "id": "gladia/solaria-3",
      "kind": "stt",
      "provider": "Gladia",
      "label": "Solaria-3",
      "local": false,
      "perMinute": 0.01016,
      "wer": 3.2,
      "speed": 63,
      "languages": 100,
      "streaming": true,
      "diarization": true,
      "note": "Accurate and very multilingual, but priced well above equally accurate rivals."
    },
    {
      "id": "groq/whisper-large-v3-turbo",
      "kind": "stt",
      "provider": "Groq",
      "label": "Whisper Large v3 Turbo",
      "local": false,
      "perMinute": 0.00067,
      "priceNote": "$0.04/hour · 10 s minimum per request",
      "wer": 4.6,
      "speed": 105,
      "languages": 99,
      "timestamps": true,
      "note": "Whisper accuracy for two-thirds of a tenth of a cent a minute. Hard to beat on bulk backfills."
    },
    {
      "id": "fal/wizper-v3",
      "kind": "stt",
      "provider": "fal.ai",
      "label": "Wizper (Whisper v3)",
      "local": false,
      "perMinute": 0.0005,
      "wer": 4.7,
      "speed": 207,
      "languages": 99,
      "note": "The cheapest cloud transcription in the table — $0.03 an hour, and fast."
    },
    {
      "id": "together/parakeet-tdt-v3",
      "kind": "stt",
      "provider": "Together AI",
      "label": "Parakeet TDT 0.6B v3",
      "local": false,
      "perMinute": 0.0015,
      "wer": 4.5,
      "speed": 893,
      "languages": 25,
      "timestamps": true,
      "note": "893× real time — the fastest thing here. An hour of audio in about four seconds."
    },
    {
      "id": "local/parakeet-tdt-v3-mlx",
      "kind": "stt",
      "provider": "NVIDIA · local",
      "label": "Parakeet TDT 0.6B v3 (MLX)",
      "local": true,
      "wer": 6.3,
      "speed": 3300,
      "languages": 25,
      "timestamps": true,
      "offline": true,
      "license": "CC-BY-4.0",
      "note": "Fastest local option on Apple Silicon, and it stays silent during silence instead of hallucinating."
    },
    {
      "id": "local/whisper-large-v3-turbo",
      "kind": "stt",
      "provider": "OpenAI · local",
      "label": "Whisper Large v3 Turbo (whisper.cpp)",
      "local": true,
      "wer": 7.4,
      "speed": 146,
      "languages": 99,
      "timestamps": true,
      "offline": true,
      "license": "MIT",
      "note": "99 languages, fully offline, free. The fallback that always works — 1.6 GB and ≥16 GB RAM."
    },
    {
      "id": "local/whisper-base",
      "kind": "stt",
      "provider": "OpenAI · local",
      "label": "Whisper Base (whisper.cpp)",
      "local": true,
      "speed": 300,
      "languages": 99,
      "timestamps": true,
      "offline": true,
      "license": "MIT",
      "note": "148 MB, runs on anything, rough on accents. The safe default on a small machine."
    },
    {
      "id": "local/voxtral-mini-mlx",
      "kind": "stt",
      "provider": "Mistral · local",
      "label": "Voxtral Mini 4B (MLX, 4-bit)",
      "local": true,
      "languages": 30,
      "offline": true,
      "license": "Apache-2.0",
      "note": "Strong on accented speech, offline. ~3 GB resident — fits 16 GB alongside a local LLM."
    },
    {
      "id": "openai/gpt-4o-mini-tts",
      "kind": "tts",
      "provider": "OpenAI",
      "label": "GPT-4o Mini TTS",
      "local": false,
      "perMChar": 15,
      "priceNote": "billed on audio tokens (~$12/1M out); ~$15/1M chars equivalent",
      "latencyMs": 300,
      "languages": 50,
      "streaming": true,
      "note": "Steerable: you can instruct the tone in plain words. 11 preset voices, no cloning."
    },
    {
      "id": "openai/tts-1",
      "kind": "tts",
      "provider": "OpenAI",
      "label": "TTS-1",
      "local": false,
      "perMChar": 15,
      "latencyMs": 400,
      "languages": 50,
      "streaming": true,
      "note": "The low-latency classic. Superseded by gpt-4o-mini-tts at the same price."
    },
    {
      "id": "openai/tts-1-hd",
      "kind": "tts",
      "provider": "OpenAI",
      "label": "TTS-1 HD",
      "local": false,
      "perMChar": 30,
      "languages": 50,
      "note": "Twice the price of tts-1 for a modest fidelity gain. For finished audio, not conversation."
    },
    {
      "id": "elevenlabs/flash-v2.5",
      "kind": "tts",
      "provider": "ElevenLabs",
      "label": "Flash v2.5",
      "local": false,
      "perMChar": 50,
      "priceNote": "$0.05/1k chars at API rates; plan credits change the effective rate",
      "latencyMs": 75,
      "languages": 32,
      "streaming": true,
      "cloning": true,
      "note": "~75 ms model latency — the ElevenLabs voice you'd put in a live conversation."
    },
    {
      "id": "elevenlabs/multilingual-v2",
      "kind": "tts",
      "provider": "ElevenLabs",
      "label": "Multilingual v2",
      "local": false,
      "perMChar": 100,
      "latencyMs": 280,
      "languages": 32,
      "streaming": true,
      "cloning": true,
      "note": "The quality benchmark most people mean by 'sounds real'. Twice Flash's price, and slower."
    },
    {
      "id": "elevenlabs/v3",
      "kind": "tts",
      "provider": "ElevenLabs",
      "label": "Eleven v3",
      "local": false,
      "perMChar": 100,
      "languages": 70,
      "cloning": true,
      "note": "Most expressive — audio tags for emotion and delivery. For narration, not low latency."
    },
    {
      "id": "xai/grok-tts",
      "kind": "tts",
      "provider": "xAI",
      "label": "Grok Voice TTS",
      "local": false,
      "perMChar": 4.2,
      "languages": 30,
      "streaming": true,
      "cloning": true,
      "note": "An order of magnitude cheaper than ElevenLabs, with cloning. The value pick for bulk speech."
    },
    {
      "id": "deepgram/aura-2",
      "kind": "tts",
      "provider": "Deepgram",
      "label": "Aura-2",
      "local": false,
      "perMChar": 30,
      "latencyMs": 150,
      "languages": 15,
      "streaming": true,
      "note": "Built for phone agents: consistent, fast, unremarkable — which is the point on a call."
    },
    {
      "id": "deepgram/aura-1",
      "kind": "tts",
      "provider": "Deepgram",
      "label": "Aura-1",
      "local": false,
      "perMChar": 15,
      "latencyMs": 130,
      "languages": 1,
      "streaming": true,
      "note": "Half Aura-2's price, English only. Fine for notifications and short confirmations."
    },
    {
      "id": "cartesia/sonic-3",
      "kind": "tts",
      "provider": "Cartesia",
      "label": "Sonic 3",
      "local": false,
      "perMChar": 35,
      "priceNote": "credit-based; ~$5–$37/1M chars depending on plan",
      "latencyMs": 90,
      "languages": 42,
      "streaming": true,
      "cloning": true,
      "note": "The latency leader — ~90 ms to first audio. What you pick when interruption has to feel instant."
    },
    {
      "id": "hume/octave-2",
      "kind": "tts",
      "provider": "Hume",
      "label": "Octave 2",
      "local": false,
      "perMChar": 7.6,
      "languages": 11,
      "streaming": true,
      "cloning": true,
      "note": "Emotion-aware delivery at a low price — it reads the sentence's feeling, not just its words."
    },
    {
      "id": "rime/mistv2",
      "kind": "tts",
      "provider": "Rime",
      "label": "Mist v2",
      "local": false,
      "perMChar": 39,
      "priceNote": "$0.030/min of audio",
      "latencyMs": 100,
      "languages": 2,
      "streaming": true,
      "note": "Deliberately casual, non-broadcast voices — the ones that don't sound like an announcer."
    },
    {
      "id": "minimax/speech-2.6",
      "kind": "tts",
      "provider": "MiniMax",
      "label": "Speech 2.6",
      "local": false,
      "perMChar": 45,
      "languages": 40,
      "streaming": true,
      "cloning": true,
      "note": "Strongest non-English coverage of the premium voices, especially across Asian languages."
    },
    {
      "id": "fishaudio/s2-pro",
      "kind": "tts",
      "provider": "Fish Audio",
      "label": "OpenAudio S2 Pro",
      "local": false,
      "perMChar": 15,
      "languages": 13,
      "cloning": true,
      "license": "open weights",
      "note": "Open weights AND a hosted API at a low rate — prototype on the API, self-host if it works out."
    },
    {
      "id": "google/chirp-3-hd",
      "kind": "tts",
      "provider": "Google Cloud",
      "label": "Chirp 3 HD",
      "local": false,
      "perMChar": 30,
      "priceNote": "first 1M chars/month free",
      "latencyMs": 500,
      "languages": 31,
      "streaming": true,
      "note": "Google's conversational tier. A free monthly million characters covers a lot of notifications."
    },
    {
      "id": "google/neural2",
      "kind": "tts",
      "provider": "Google Cloud",
      "label": "Neural2",
      "local": false,
      "perMChar": 16,
      "languages": 40,
      "note": "The workhorse Google voice. Half the HD price and perfectly serviceable for read-aloud."
    },
    {
      "id": "azure/neural-hd",
      "kind": "tts",
      "provider": "Microsoft Azure",
      "label": "Neural HD",
      "local": false,
      "perMChar": 22,
      "priceNote": "cut from $30 to $22/1M in March 2026; commitment tiers reach ~$7.50/1M",
      "latencyMs": 300,
      "languages": 140,
      "streaming": true,
      "note": "140 languages — the broadest TTS coverage. Commitment pricing beats everyone at volume."
    },
    {
      "id": "azure/neural",
      "kind": "tts",
      "provider": "Microsoft Azure",
      "label": "Neural",
      "local": false,
      "perMChar": 16,
      "latencyMs": 300,
      "languages": 140,
      "streaming": true,
      "note": "Same reach as Neural HD, plainer delivery, cheaper."
    },
    {
      "id": "aws/polly-generative",
      "kind": "tts",
      "provider": "AWS",
      "label": "Polly Generative",
      "local": false,
      "perMChar": 30,
      "languages": 12,
      "streaming": true,
      "note": "Polly's conversational tier. Fine, and it bills on the account you already have."
    },
    {
      "id": "aws/polly-neural",
      "kind": "tts",
      "provider": "AWS",
      "label": "Polly Neural",
      "local": false,
      "perMChar": 16,
      "latencyMs": 400,
      "languages": 40,
      "streaming": true,
      "note": "Cheap, fast, dependable, and it sounds like 2021. Good for IVR, not for a host voice."
    },
    {
      "id": "aws/polly-standard",
      "kind": "tts",
      "provider": "AWS",
      "label": "Polly Standard",
      "local": false,
      "perMChar": 4,
      "latencyMs": 300,
      "languages": 29,
      "streaming": true,
      "note": "The cheapest cloud TTS anywhere, and audibly robotic. Use where nobody's listening closely."
    },
    {
      "id": "local/kokoro",
      "kind": "tts",
      "provider": "Kokoro · local",
      "label": "Kokoro 82M",
      "local": true,
      "latencyMs": 120,
      "languages": 8,
      "offline": true,
      "license": "Apache-2.0",
      "note": "82M params, near-cloud quality, runs on a laptop CPU. The best free voice — needs its own server on :8880."
    },
    {
      "id": "local/piper",
      "kind": "tts",
      "provider": "Piper · local",
      "label": "Piper",
      "local": true,
      "latencyMs": 80,
      "languages": 30,
      "offline": true,
      "license": "MIT",
      "note": "Offline neural voices, instant, tiny. Clearly better than the system voice and still free."
    },
    {
      "id": "local/say",
      "kind": "tts",
      "provider": "Apple · local",
      "label": "macOS say (built-in)",
      "local": true,
      "latencyMs": 50,
      "languages": 40,
      "offline": true,
      "note": "Zero setup, dozens of system voices, works on a plane. Dated, but it never fails."
    },
    {
      "id": "local/chatterbox",
      "kind": "tts",
      "provider": "Resemble · local",
      "label": "Chatterbox Turbo",
      "local": true,
      "latencyMs": 200,
      "languages": 23,
      "cloning": true,
      "offline": true,
      "license": "MIT",
      "note": "Open-weight voice cloning that beat ElevenLabs in blind tests. Needs a GPU; no adapter yet."
    },
    {
      "id": "openai/gpt-realtime-2.1",
      "kind": "realtime",
      "provider": "OpenAI",
      "label": "GPT Realtime 2.1",
      "local": false,
      "perMAudioToken": {
        "in": 32,
        "out": 64,
        "tokensPerSecIn": 10,
        "tokensPerSecOut": 10
      },
      "priceNote": "audio tokens: $32/1M in, $64/1M out (~10 tokens/s of audio)",
      "latencyMs": 500,
      "languages": 60,
      "streaming": true,
      "note": "The most capable speech-to-speech: tools, interruption, tone. Also the most expensive per minute."
    },
    {
      "id": "openai/gpt-realtime-mini",
      "kind": "realtime",
      "provider": "OpenAI",
      "label": "GPT Realtime Mini",
      "local": false,
      "perMAudioToken": {
        "in": 10,
        "out": 20,
        "tokensPerSecIn": 10,
        "tokensPerSecOut": 10
      },
      "priceNote": "audio tokens: $10/1M in, $20/1M out",
      "latencyMs": 450,
      "languages": 60,
      "streaming": true,
      "note": "A third of the flagship's price with the same protocol. The sensible realtime default."
    },
    {
      "id": "google/gemini-live",
      "kind": "realtime",
      "provider": "Google",
      "label": "Gemini Live",
      "local": false,
      "perMAudioToken": {
        "in": 3,
        "out": 12,
        "tokensPerSecIn": 32,
        "tokensPerSecOut": 25
      },
      "priceNote": "audio: $3/1M in (32 tok/s), $12/1M out (25 tok/s)",
      "latencyMs": 600,
      "languages": 70,
      "streaming": true,
      "note": "Handles interruptions best, and it can see — screen or camera in the same session."
    },
    {
      "id": "xai/grok-voice-agent",
      "kind": "realtime",
      "provider": "xAI",
      "label": "Grok Voice Agent",
      "local": false,
      "perMinute": 0.05,
      "priceNote": "flat per minute of audio, voices included; +$0.01/min with a provisioned number",
      "latencyMs": 500,
      "languages": 30,
      "streaming": true,
      "note": "Flat per-minute billing — no token arithmetic. Cheapest full speech-to-speech with personality."
    },
    {
      "id": "elevenlabs/speech-engine",
      "kind": "realtime",
      "provider": "ElevenLabs",
      "label": "Speech Engine",
      "local": false,
      "perMinute": 0.08,
      "languages": 90,
      "streaming": true,
      "cloning": true,
      "note": "ElevenLabs voice quality end-to-end, at ElevenLabs prices."
    },
    {
      "id": "deepgram/voice-agent",
      "kind": "realtime",
      "provider": "Deepgram",
      "label": "Voice Agent API",
      "local": false,
      "perMinute": 0.075,
      "priceNote": "standard tier; bring-your-own LLM/TTS costs less",
      "languages": 36,
      "streaming": true,
      "note": "One socket for the whole loop, tuned for telephony. Swap in your own LLM to cut the rate."
    },
    {
      "id": "cartesia/line",
      "kind": "realtime",
      "provider": "Cartesia",
      "label": "Line",
      "local": false,
      "perMinute": 0.06,
      "priceNote": "+$0.014/min with a Cartesia phone number",
      "latencyMs": 90,
      "languages": 42,
      "streaming": true,
      "cloning": true,
      "note": "Sonic's latency in a full agent loop — the one that feels most like a real phone call."
    }
  ],
  "fal": [
    {
      "id": "bytedance/seedance-2.5/flash/image-to-video",
      "title": "Seedance 2.5 Flash Image to Video",
      "category": "image-to-video",
      "family": "Seedance 2.5",
      "description": "Dreamina Seedance 2.5 Flash is the fast tier of Seedance 2.5. It animates a still image into a 4 to 30 second clip with native audio at 480p or 720p, with optional end-frame control.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab285d2/MHFgbC6FvdXJB56-miePl.jpg",
      "price": "For 480p, your request will cost roughly $0.1442 per second of generated video, and for 720p, you will be charged roughly $0.3094 per second of generated video. Your request will cost $0.014 per 1,000 tokens. The number of tokens is roughly",
      "license": "commercial",
      "published": "2026-11-15",
      "url": "https://fal.ai/models/bytedance/seedance-2.5/flash/image-to-video"
    },
    {
      "id": "bytedance/seedance-2.5/flash/reference-to-video",
      "title": "Seedance 2.5 Flash Reference to Video",
      "category": "image-to-video",
      "family": "Seedance 2.5",
      "description": "Dreamina Seedance 2.5 Flash is the fast tier of Seedance 2.5. It generates, edits and extends video from up to 10 images, 5 videos and 5 audio references at 480p or 720p.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab285d5/UZkA73-Vbq1dyWpa4QuDK.jpg",
      "price": "For 480p, your request will cost roughly $0.1442 per second of generated video, and for 720p, you will be charged roughly $0.3094 per second of generated video. Your request will cost $0.014 per 1,000 tokens. The number of tokens is roughly",
      "license": "commercial",
      "published": "2026-11-15",
      "url": "https://fal.ai/models/bytedance/seedance-2.5/flash/reference-to-video"
    },
    {
      "id": "bytedance/seedance-2.5/flash/text-to-video",
      "title": "Seedance 2.5 Flash Text to Video",
      "category": "text-to-video",
      "family": "Seedance 2.5",
      "description": "Dreamina Seedance 2.5 Flash is the fast tier of Seedance 2.5. It generates 4 to 30 second video with native audio from a text prompt at 480p or 720p.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab285cf/e6awRuthfFlNdoFdftBz0.jpg",
      "price": "For 480p, your request will cost roughly $0.1442 per second of generated video, and for 720p, you will be charged roughly $0.3094 per second of generated video. Your request will cost $0.014 per 1,000 tokens. The number of tokens is roughly",
      "license": "commercial",
      "published": "2026-11-15",
      "url": "https://fal.ai/models/bytedance/seedance-2.5/flash/text-to-video"
    },
    {
      "id": "bytedance/seedream/v5/pro/us/edit",
      "title": "Seedream 5.0 Pro US",
      "category": "image-to-image",
      "family": "Seedream 5.0 Pro",
      "description": "US inference endpoint for Seedream 5.0 Pro Image Editing.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab27ad8/n1tuY9R2D6M2Drik0MWUT.jpg",
      "price": "You will be charged for both input and output images. The first input image is not charged, and every additional input image will cost $0.0045. Tentative pricing is $0.0675+ $(0.0045 x number of additional input images) per output image for",
      "license": "commercial",
      "published": "2026-11-14",
      "url": "https://fal.ai/models/bytedance/seedream/v5/pro/us/edit"
    },
    {
      "id": "bytedance/seedream/v5/pro/us/text-to-image",
      "title": "Seedream",
      "category": "text-to-image",
      "family": "Seedream 5",
      "description": "US Inference endpoint for Seedream 5.0 Pro Text-to-Image.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab27a89/tJ8nKYhMXBflAnFhUl7Py.jpg",
      "price": "Pricing is $0.0675 per output image for an area up to and including 1536 × 1536 pixels, or $0.135 per output image for an area above 1536 × 1536 and up to 2048 × 2048 pixels. For example, 2 images at 1536 × 1536 would cost $0.135.",
      "license": "commercial",
      "published": "2026-11-14",
      "url": "https://fal.ai/models/bytedance/seedream/v5/pro/us/text-to-image"
    },
    {
      "id": "bytedance/seedream/v5.5/edit",
      "title": "Seedream 5.5 Edit",
      "category": "image-to-image",
      "family": "Seedream 5.5",
      "description": "Seedream 5.5 edits images from text instructions and up to 10 reference images, with configurable output dimensions and support for transparent PNG backgrounds.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab20610/46iJg1tOFbk6HNZ3Rs9hX.jpg",
      "price": "Billing is calculated per output image, including an input-image fee for each output generated. Each output costs $0.03 at 1K or 1.5K, or $0.0405 at 2K, plus $0.012 per input image per output. For custom dimensions, the output rate is $0.03",
      "license": "commercial",
      "published": "2026-11-12",
      "url": "https://fal.ai/models/bytedance/seedream/v5.5/edit"
    },
    {
      "id": "bytedance/seedream/v5.5/layerize",
      "title": "Seedream 5.5 Layerize",
      "category": "image-to-image",
      "family": "Seedream 5.5",
      "description": "Seedream 5.5 decomposes an image into a base image and independently editable PNG layers, with layer positions, stacking order, and optional text recognition.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab2061b/gTIQ-HHJnFpDBUdz5zref.jpg",
      "price": "Billing is calculated per returned output image, including the base image and every separated layer, plus $0.012 for the input image once per request. Each output costs $0.03 at 1K or 1.5K, or $0.0405 at 2K. With auto resolution, the base i",
      "license": "commercial",
      "published": "2026-11-12",
      "url": "https://fal.ai/models/bytedance/seedream/v5.5/layerize"
    },
    {
      "id": "bytedance/seedream/v5.5/text-to-image",
      "title": "Seedream 5.5 Text to Image",
      "category": "text-to-image",
      "family": "Seedream 5.5",
      "description": "Seedream 5.5 generates images from text prompts with configurable dimensions, batch generation, and support for transparent PNG backgrounds.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab20606/Qo42-iL_Eu9QCiX31ZTh-.jpg",
      "price": "Billing is calculated per output image. Each image costs $0.03 at 1K or 1.5K, or $0.0405 at 2K. For custom dimensions, the rate is $0.03 up to and including 2,610,000 pixels, and $0.0405 above that area. For example, 2 output images cost $0",
      "license": "commercial",
      "published": "2026-11-12",
      "url": "https://fal.ai/models/bytedance/seedream/v5.5/text-to-image"
    },
    {
      "id": "pixverse/v6/max/image-to-video",
      "title": "PixVerse V6 Max Image To Video",
      "category": "image-to-video",
      "family": "Pixverse 6",
      "description": "Generate videos from images with PixVerse V6 Max, with optional audio and multi-clip generation.",
      "thumbnail": "https://v3b.fal.media/files/b/0a94048a/NbqGsBzItNNlvMmlYie7E_7d241b8764b64124b42edbcd5da51960.jpg",
      "price": "Billing is calculated per second of video generated. For 360p, your request will cost $0.054 per second without audio and $0.081 with audio; for 540p, $0.081 without audio and $0.099 with audio; for 720p, $0.099 without audio and $0.135 wit",
      "license": "commercial",
      "published": "2026-11-11",
      "url": "https://fal.ai/models/pixverse/v6/max/image-to-video"
    },
    {
      "id": "pixverse/v6/max/text-to-video",
      "title": "PixVerse V6 Max Text To Video",
      "category": "text-to-video",
      "family": "pixverse-v6-max",
      "description": "Generate videos from text with PixVerse V6 Max, with optional audio and multi-clip generation.",
      "thumbnail": "https://v3b.fal.media/files/b/0a940489/sEtg0Kth_fWhHMR5MS664_222abe748d96461ba565ba860c3061d8.jpg",
      "price": "Billing is calculated per second of video generated. For 360p, your request will cost $0.054 per second without audio and $0.081 with audio; for 540p, $0.081 without audio and $0.099 with audio; for 720p, $0.099 without audio and $0.135 wit",
      "license": "commercial",
      "published": "2026-11-11",
      "url": "https://fal.ai/models/pixverse/v6/max/text-to-video"
    },
    {
      "id": "lightricks/ltx-2.5/alpha-gen",
      "title": "LTX 2.5 Alpha Gen",
      "category": "video-to-video",
      "family": "LTX 2.5",
      "description": "LTX 2.5 Alpha Gen extracts frame-matched foreground mattes from existing video, preserving partial transparency in hair and soft edges. It returns a grayscale matte MP4 and a ProRes 4444 cutout with straight alpha for compositing.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1efe3/xbbXyiVUb3gOnkbEYcBdl.jpg",
      "price": "Billing is calculated per second of input video, based on its resolution: $0.10 at 1080p, $0.25 at 1440p, and $1.00 at 2160p. A 5-second input costs $0.50, $1.25, or $5.00, respectively. The tier is determined by the input video’s shorter s",
      "license": "commercial",
      "published": "2026-11-10",
      "url": "https://fal.ai/models/lightricks/ltx-2.5/alpha-gen"
    },
    {
      "id": "lightricks/ltx-2.5/hdr",
      "title": "LTX 2.5 SDR to HDR",
      "category": "video-to-video",
      "family": "LTX 2.5",
      "description": "LTX 2.5 SDR to HDR converts existing SDR video for HDR finishing workflows. It returns a processed video and a ZIP archive of 16-bit half-float OpenEXR frames, with selectable ACEScg, ACEScct, linear sRGB, or ACES2065-1 color space.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1effa/NZjci-MJMqtQUa3W0EBf5.jpg",
      "price": "Billing is calculated per second of input video, based on its resolution: $0.20 at 1080p, $0.40 at 1440p, and $0.80 at 2160p. A 5-second input costs $1.00, $2.00, or $4.00, respectively. The tier is determined by the input pixel count: up t",
      "license": "commercial",
      "published": "2026-11-10",
      "url": "https://fal.ai/models/lightricks/ltx-2.5/hdr"
    },
    {
      "id": "lightricks/ltx-2.5/reframe",
      "title": "LTX 2.5 Reframe",
      "category": "video-to-video",
      "family": "LTX 2.5",
      "description": "LTX 2.5 Reframe adapts existing video to a new aspect ratio by generating missing frame content. It supports five aspect ratios, 720p and 1080p output, and source placement controls for fitting and positioning the original video.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1efeb/UUBR2Tp20ZDOzijQuDVmY.jpg",
      "price": "Billing is calculated per second of input video, based on the output resolution: $0.10 at 720p and $0.20 at 1080p. A 5-second input costs $0.50 at 720p or $1.00 at 1080p.",
      "license": "commercial",
      "published": "2026-11-10",
      "url": "https://fal.ai/models/lightricks/ltx-2.5/reframe"
    },
    {
      "id": "lightricks/ltx-2.5/upscale",
      "title": "LTX 2.5 Upscale",
      "category": "video-to-video",
      "family": "LTX 2.5",
      "description": "LTX 2.5 Upscale increases the resolution of existing H.264 or HEVC video to 1080p, 4K, or 8K. It preserves the source aspect ratio, retains audio in stereo, and rounds the source frame rate to an integer.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1eff4/I9Qua7Bg5vc1oLhw__j9S.jpg",
      "price": "Billing is calculated per second of input video, based on the output resolution: $0.08 at 1080p, $0.30 at 4K, and $1.20 at 8K. A 5-second input costs $0.40, $1.50, or $6.00, respectively.",
      "license": "commercial",
      "published": "2026-11-10",
      "url": "https://fal.ai/models/lightricks/ltx-2.5/upscale"
    },
    {
      "id": "minimax/h3-image/edit",
      "title": "MiniMax H3 Image Edit",
      "category": "image-to-image",
      "family": "H3 Image",
      "description": "MiniMax H3 Image Edit transforms and combines up to 16 reference images using text instructions, supporting targeted edits, style changes, and multi-image composition. Its unified generation and editing model produces detailed results at custom output sizes up to 20 megapixels.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1e28c/Dl3tlJm8lzFzM3wPL4xaj.jpg",
      "price": "Billed on MiniMax token usage: $2.30 per 1M input text tokens, $1.50 per 1M input image tokens and $3.80 per 1M output image tokens.",
      "license": "commercial",
      "published": "2026-11-10",
      "url": "https://fal.ai/models/minimax/h3-image/edit"
    },
    {
      "id": "minimax/h3-image/highspeed/edit",
      "title": "MiniMax H3 Image Highspeed Edit",
      "category": "image-to-image",
      "family": "H3 Image",
      "description": "MiniMax H3 Image Highspeed Edit transforms and combines up to 16 reference images using text instructions. It supports multi-image composition and custom output sizes up to 20 megapixels, preserving the first image’s aspect ratio by default.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1ec6e/dPJpcuGdqvGAOvQNFr0m4.jpg",
      "price": "Billing is based on actual token usage: $23 per 10 million input text tokens, $30 per 10 million input image tokens, and $76 per 10 million output image tokens. Prompt enhancement is included in reported input text token usage.",
      "license": "commercial",
      "published": "2026-11-10",
      "url": "https://fal.ai/models/minimax/h3-image/highspeed/edit"
    },
    {
      "id": "minimax/h3-image/highspeed/text-to-image",
      "title": "MiniMax H3 Image Highspeed Text to Image",
      "category": "text-to-image",
      "family": "H3 Image",
      "description": "MiniMax H3 Image Highspeed generates detailed images from text prompts using the Highspeed variant of MiniMax H3 Image. It supports custom output sizes up to 20 megapixels and optional prompt expansion that develops short prompts into complete image descriptions.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1ec68/hli5UyIAw77t0-wrPSvzg.jpg",
      "price": "Billing is based on actual token usage: $23 per 10 million input text tokens and $76 per 10 million output image tokens. Prompt enhancement is included in reported input text token usage.",
      "license": "commercial",
      "published": "2026-11-10",
      "url": "https://fal.ai/models/minimax/h3-image/highspeed/text-to-image"
    },
    {
      "id": "minimax/h3-image/text-to-image",
      "title": "MiniMax H3 Image Text to Image",
      "category": "text-to-image",
      "family": "H3 Image",
      "description": "MiniMax H3 Image generates detailed images from text prompts, with precise instruction following across photorealistic scenes, illustrations, and visual designs. It supports native high-resolution generation up to 20 megapixels and optional prompt expansion that develops short prompts into complete ",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1e27d/wIPqNN4tTDZ-xkr7b7L_l.jpg",
      "price": "Billed on MiniMax token usage: $2.30 per 1M input text tokens and $3.80 per 1M output image tokens.",
      "license": "commercial",
      "published": "2026-11-10",
      "url": "https://fal.ai/models/minimax/h3-image/text-to-image"
    },
    {
      "id": "openimaginglab/flashvsr-plus",
      "title": "FlashVSR+",
      "category": "video-to-video",
      "family": "FlashVSR",
      "description": "FlashVSR+ upscales video by up to 4x with a one-step diffusion model, offering a fast Turbo mode and a higher-quality Base mode. Output reaches about 4K, and the input audio track is preserved by default.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab13a7e/Xtaet5vjuOL41xyspNr0i.jpg",
      "price": "Your request will cost $0.0005 per megapixel of upscaled video (output width × height × frames). For example, upscaling a 10-second 480p clip by 4x to 3416×1920 (241 frames) costs $0.80, and a 1920×1080 output with 121 frames costs $0.13. S",
      "license": "commercial",
      "published": "2026-11-10",
      "url": "https://fal.ai/models/openimaginglab/flashvsr-plus"
    },
    {
      "id": "meshy/t2.5/image-to-3d",
      "title": "Meshy T2.5 Image to 3D",
      "category": "image-to-3d",
      "family": "T2.5",
      "description": "Meshy T2.5 is Meshy's latest Smart Topology model. It converts a single image into a clean, triangle-only 3D mesh with natively separated parts, with a target polycount of up to 30,000 triangles (default 15,000). Compared with Meshy T2, it produces sharper detail and keeps shape and proportions clos",
      "thumbnail": "https://v3b.fal.media/files/b/0ab16b7a/BBJ6vAcsczbyPk6Zc3x4B.jpg",
      "price": "Billing is calculated per generation. An untextured mesh costs $0.20, and a textured mesh costs $0.60; PBR maps and the choice of texture model are included at no extra charge. Optional auto-rigging adds $0.20, and animation adds $0.12 per ",
      "license": "commercial",
      "published": "2026-11-06",
      "url": "https://fal.ai/models/meshy/t2.5/image-to-3d"
    },
    {
      "id": "meshy/t2.5/text-to-3d",
      "title": "Meshy T2.5 Text to 3D",
      "category": "text-to-3d",
      "family": "T2.5",
      "description": "Meshy T2.5 is Meshy's latest Smart Topology model. It generates a clean, triangle-only 3D mesh with natively separated parts from a text prompt, with a target polycount of up to 30,000 triangles (default 15,000). Preview mode returns the untextured mesh, and full mode textures it with Meshy 7.1; com",
      "thumbnail": "https://v3b.fal.media/files/b/0ab16c19/jQs2tHoLUvyaL598htsIH.jpg",
      "price": "Billing is calculated per generation. Preview mode, which returns the untextured mesh, costs $0.20, and full mode, which also textures it, costs $0.60; PBR maps and the choice of texture model are included at no extra charge. Optional auto-",
      "license": "commercial",
      "published": "2026-11-06",
      "url": "https://fal.ai/models/meshy/t2.5/text-to-3d"
    },
    {
      "id": "meshy/t2/image-to-3d",
      "title": "Meshy T2 Image to 3D",
      "category": "image-to-3d",
      "family": "T2",
      "description": "Meshy T2 is Meshy's faster Smart Topology model. It converts a single image into a clean, triangle-only 3D mesh with natively separated parts, with a target polycount of up to 15,000 triangles (default 4,000). It generates faster than Meshy T2.5, which suits low-poly assets and speed-sensitive pipel",
      "thumbnail": "https://v3b.fal.media/files/b/0ab16c98/QtFIM0dcuBYWoCzXB50Bi.jpg",
      "price": "Billing is calculated per generation. An untextured mesh costs $0.20, and a textured mesh costs $0.60; PBR maps and the choice of texture model are included at no extra charge. Optional auto-rigging adds $0.20, and animation adds $0.12 per ",
      "license": "commercial",
      "published": "2026-11-06",
      "url": "https://fal.ai/models/meshy/t2/image-to-3d"
    },
    {
      "id": "meshy/t2/text-to-3d",
      "title": "Meshy T2 Text to 3D",
      "category": "text-to-3d",
      "family": "T2",
      "description": "Meshy T2 is Meshy's faster Smart Topology model. It generates a clean, triangle-only 3D mesh with natively separated parts from a text prompt, with a target polycount of up to 15,000 triangles (default 4,000). Preview mode returns the untextured mesh, and full mode textures it with Meshy 7.1; it gen",
      "thumbnail": "https://v3b.fal.media/files/b/0ab16c69/VKU-mA8WVdW6FxDC7Eheq.jpg",
      "price": "Billing is calculated per generation. Preview mode, which returns the untextured mesh, costs $0.20, and full mode, which also textures it, costs $0.60; PBR maps and the choice of texture model are included at no extra charge. Optional auto-",
      "license": "commercial",
      "published": "2026-11-06",
      "url": "https://fal.ai/models/meshy/t2/text-to-3d"
    },
    {
      "id": "microsoft/moge-2",
      "title": "MoGe-2",
      "category": "image-to-3d",
      "family": "MoGe",
      "description": "Estimates metric 3D geometry from a single image, returning a point map, depth, a validity mask and the camera field of view, with surface normals on the normal-capable checkpoints. Four ViT checkpoints trade accuracy for speed, and the result can be exported as a textured GLB mesh and a colored PLY",
      "thumbnail": "https://v3b.fal.media/files/b/0ab14209/abGVDqNNAJSfTGkPDY7Yj.jpg",
      "price": "Billing is calculated per megapixel of the input image, rounded up to the next whole megapixel, with a minimum of 1 megapixel per request. Each megapixel costs $0.01. A 1600×1067 image (1.7 MP) costs $0.02, a 1920×1080 image (2.1 MP) costs ",
      "license": "commercial",
      "published": "2026-11-06",
      "url": "https://fal.ai/models/microsoft/moge-2"
    },
    {
      "id": "microsoft/moge-3",
      "title": "MoGe-3",
      "category": "image-to-3d",
      "family": "MoGe",
      "description": "Estimates metric 3D geometry from a single image, returning a point map, depth, surface normals, a validity mask and the camera field of view. A sparse 3D refiner sharpens the predicted depth, and the result can be exported as a textured GLB mesh and a colored PLY point cloud.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab141ee/YUzWPTOTUIOGTc_qU-n0m.jpg",
      "price": "Billing is calculated per megapixel of the input image, rounded up to the next whole megapixel, with a minimum of 1 megapixel per request. Each megapixel costs $0.015. A 1600×1067 image (1.7 MP) costs $0.03, a 1920×1080 image (2.1 MP) costs",
      "license": "commercial",
      "published": "2026-11-06",
      "url": "https://fal.ai/models/microsoft/moge-3"
    },
    {
      "id": "decart/lucy-2-6/realtime",
      "title": "Lucy 2 6",
      "category": "video-to-video",
      "family": "Lucy",
      "description": "Real-time, prompt-driven video editing over WebRTC. Restyle, swap backgrounds, and add or replace objects live on a webcam or streamed feed at interactive latency.",
      "thumbnail": "https://v3b.fal.media/files/b/0aa25d26/n0Ml0BWna6ulj0yHPm4gj.jpg",
      "license": "commercial",
      "published": "2026-11-05",
      "url": "https://fal.ai/models/decart/lucy-2-6/realtime"
    },
    {
      "id": "google/gemini-omni-flash/v1.2/edit",
      "title": "Gemini Omni Flash 1.2 Edit Video",
      "category": "video-to-video",
      "family": "Omni Flash 1.2",
      "description": "Gemini Omni Flash 1.2 is Google's multimodal video model. This endpoint edits an existing video of up to 15 seconds from a natural-language instruction, applying the requested change while keeping the rest of the scene unchanged. Output resolution is selectable from 360p, 720p, 1080p, or 4K.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab147a4/Ah1aEWi0zcfCcuBJra3JK.jpg",
      "price": "Billing is calculated per second of the edited output video, by resolution. For 360p, your request will cost $0.0273 per second; for 720p, $0.0819 per second; for 1080p, $0.1229 per second; and for 4K, $0.2457 per second. Editing a 10-secon",
      "license": "commercial",
      "published": "2026-11-05",
      "url": "https://fal.ai/models/google/gemini-omni-flash/v1.2/edit"
    },
    {
      "id": "google/gemini-omni-flash/v1.2/extend",
      "title": "Gemini Omni Flash 1.2 Extend Video",
      "category": "video-to-video",
      "family": "Omni Flash 1.2",
      "description": "Gemini Omni Flash 1.2 is Google's multimodal video model. This endpoint continues an existing video from a text prompt, conditioning on its last 10 seconds and appending a new 3 to 10 second segment. Output resolution is selectable from 360p, 720p, 1080p, or 4K.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab147ad/3B0cVqqGKQFPABIe-Qw_d.jpg",
      "price": "Billing is calculated per second of newly generated video, by resolution; the input video is not billed. For 360p, your request will cost $0.0273 per second; for 720p, $0.0819 per second; for 1080p, $0.1229 per second; and for 4K, $0.2457 p",
      "license": "commercial",
      "published": "2026-11-05",
      "url": "https://fal.ai/models/google/gemini-omni-flash/v1.2/extend"
    },
    {
      "id": "google/gemini-omni-flash/v1.2/image-to-video",
      "title": "Gemini Omni Flash 1.2 Image to Video",
      "category": "image-to-video",
      "family": "Omni Flash 1.2",
      "description": "Gemini Omni Flash 1.2 is Google's multimodal video model. This endpoint animates a start image into video with synchronized native audio, with an optional end image for first-to-last-frame interpolation of up to 10 seconds. Output is available at 360p, 720p, 1080p, or 4K, with durations from 3 to 15",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1478e/Nq_aSIbp7BGc5iHqOg3X4.jpg",
      "price": "Billing is calculated per second of output video, by resolution. For 360p, your request will cost $0.0273 per second; for 720p, $0.0819 per second; for 1080p, $0.1229 per second; and for 4K, $0.2457 per second. Native audio is included, and",
      "license": "commercial",
      "published": "2026-11-05",
      "url": "https://fal.ai/models/google/gemini-omni-flash/v1.2/image-to-video"
    },
    {
      "id": "google/gemini-omni-flash/v1.2/reference-to-video",
      "title": "Gemini Omni Flash 1.2 Reference to Video",
      "category": "image-to-video",
      "family": "Omni Flash 1.2",
      "description": "Gemini Omni Flash 1.2 is Google's multimodal video model. This endpoint generates video from a text prompt combined with up to ten reference images and up to three reference videos, each addressable in the prompt, using them as visual references for subjects and settings. Output is available at 360p",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1479a/ZRG079SdIfk3mwVWCw4l_.jpg",
      "price": "Billing is calculated per second of output video, by resolution. For 360p, your request will cost $0.0273 per second; for 720p, $0.0819 per second; for 1080p, $0.1229 per second; and for 4K, $0.2457 per second. Native output audio is includ",
      "license": "commercial",
      "published": "2026-11-05",
      "url": "https://fal.ai/models/google/gemini-omni-flash/v1.2/reference-to-video"
    },
    {
      "id": "google/gemini-omni-flash/v1.2/text-to-video",
      "title": "Gemini Omni Flash 1.2 Text to Video",
      "category": "text-to-video",
      "family": "Omni Flash 1.2",
      "description": "Gemini Omni Flash 1.2 is Google's multimodal video model. This endpoint generates video with synchronized native audio from a text prompt, at 360p, 720p, 1080p, or 4K, in 16:9 or 9:16, with durations from 3 to 15 seconds.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab14781/UISxw0DDUlcr-cybhArB1.jpg",
      "price": "Billing is calculated per second of output video, by resolution. For 360p, your request will cost $0.0273 per second; for 720p, $0.0819 per second; for 1080p, $0.1229 per second; and for 4K, $0.2457 per second. Native audio is included. A 1",
      "license": "commercial",
      "published": "2026-11-05",
      "url": "https://fal.ai/models/google/gemini-omni-flash/v1.2/text-to-video"
    },
    {
      "id": "google/gemini-omni-flash/v1.2/upscale",
      "title": "Gemini Omni Flash 1.2 Upscale Video",
      "category": "video-to-video",
      "family": "Omni Flash 1.2",
      "description": "Gemini Omni Flash 1.2 is Google's multimodal video model. This endpoint upscales a video of up to 15 seconds to a selected resolution of up to 4K, with an optional text instruction to guide the result.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab147b8/KMlDjgQp9UaUoPfYde2ND.jpg",
      "price": "Billing is calculated per second of upscaled output video, based on the source and target resolution. Upscaling 360p to 720p or 720p to 1080p costs about $0.041 per second; 360p to 1080p costs $0.0819 per second; and 720p or 1080p to 4K cos",
      "license": "commercial",
      "published": "2026-11-05",
      "url": "https://fal.ai/models/google/gemini-omni-flash/v1.2/upscale"
    },
    {
      "id": "meshy/relief",
      "title": "Meshy Relief",
      "category": "image-to-3d",
      "description": "Meshy Relief converts an image into a 3D relief model with a raised surface and a flat back.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab14712/O_pzDwImb_keY9qMKpo7O.jpg",
      "price": "Your request will cost $0.80 per generation.",
      "license": "commercial",
      "published": "2026-11-05",
      "url": "https://fal.ai/models/meshy/relief"
    },
    {
      "id": "kling/v4/flash/image-to-video",
      "title": "Kling 4.0 Flash Image to Video",
      "category": "image-to-video",
      "family": "Kling v4 Flash",
      "description": "Kling 4.0 Flash animates a starting image into a 720p video, with optional text prompts and durations from 3 to 20 seconds.",
      "thumbnail": "https://v3b.fal.media/files/b/0aac440b/ukRC-4FsGtgEuh-uVXJAK.jpg",
      "price": "Video costs $0.094 per second at 720p.",
      "license": "commercial",
      "published": "2026-11-04",
      "url": "https://fal.ai/models/kling/v4/flash/image-to-video"
    },
    {
      "id": "kling/v4/flash/reference-to-video",
      "title": "Kling 4.0 Flash Reference to Video",
      "category": "video-to-video",
      "family": "Kling v4 Flash",
      "description": "Kling 4.0 Flash generates 720p videos from text prompts and image or video references, with durations from 3 to 20 seconds.",
      "thumbnail": "https://v3b.fal.media/files/b/0aac4418/wEurA8fFrSmy6rP6jqYpn.jpg",
      "price": "Video costs $0.094 per second at 720p. With video references, it costs $0.063 per second, billed on the output duration plus the greater of the input video duration and the output duration.",
      "license": "commercial",
      "published": "2026-11-04",
      "url": "https://fal.ai/models/kling/v4/flash/reference-to-video"
    },
    {
      "id": "kling/v4/flash/text-to-video",
      "title": "Kling 4.0 Flash Text to Video",
      "category": "text-to-video",
      "family": "Kling v4 Flash",
      "description": "Kling 4.0 Flash generates videos from text prompts at 720p, with durations from 3 to 20 seconds and configurable aspect ratios.",
      "thumbnail": "https://v3b.fal.media/files/b/0aac43e9/RDST_nZ6cCgMOlimfd3LJ.jpg",
      "price": "Video costs $0.094 per second at 720p.",
      "license": "commercial",
      "published": "2026-11-04",
      "url": "https://fal.ai/models/kling/v4/flash/text-to-video"
    },
    {
      "id": "kling/v4/flash/video-to-video",
      "title": "Kling 4.0 Flash Video Editing",
      "category": "video-to-video",
      "family": "v4",
      "description": "Kling 4.0 Flash edits an input video using text instructions and optional image, video or element references, producing 3 to 20-second videos at 720p.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab120bb/w0HVopjuQHDaXLX9I-pfa.jpg",
      "price": "Video costs $0.063 per second at 720p, billed on the output duration plus the greater of the input video duration and the output duration.",
      "license": "commercial",
      "published": "2026-11-04",
      "url": "https://fal.ai/models/kling/v4/flash/video-to-video"
    },
    {
      "id": "openai/gpt-image-2.6/flare/edit",
      "title": "GPT Image 2.6 Flare Edit",
      "category": "image-to-image",
      "family": "GPT Image 2.6",
      "description": "Precise image editing that changes only what's asked, keeping subject, composition, and background intact, with reference subjects staying recognizable across styles and successive edits.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab12545/nBSD1qllETGJZIIT84YeH.jpg",
      "price": "Text tokens (per 1M): $5.00 input, $1.25 cached, $10.00 output. Image tokens (per 1M): $8.00 input, $2.00 cached, $30.00 output. Changing the quality parameter significantly affects cost; by default we use high. Adjust it to your preference",
      "license": "commercial",
      "published": "2026-11-04",
      "url": "https://fal.ai/models/openai/gpt-image-2.6/flare/edit"
    },
    {
      "id": "openai/gpt-image-2.6/flare/text-to-image",
      "title": "GPT Image 2.6 Flare Text to Image",
      "category": "text-to-image",
      "family": "GPT Image 2.6",
      "description": "OpenAI's default image model for most applications. Fast, high-quality generation with natural lighting, rich textures, and support for complex layouts including transparent backgrounds.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab12535/2l8URtrpYsFXNVNdKhSOG.jpg",
      "price": "Text tokens (per 1M): $5.00 input, $1.25 cached, $10.00 output. Image tokens (per 1M): $8.00 input, $2.00 cached, $30.00 output. Changing the quality parameter significantly affects cost; by default we use high. Adjust it to your preference",
      "license": "commercial",
      "published": "2026-11-04",
      "url": "https://fal.ai/models/openai/gpt-image-2.6/flare/text-to-image"
    },
    {
      "id": "openai/gpt-image-2.6/sunburst/edit",
      "title": "GPT Image 2.6 Sunburst Edit",
      "category": "image-to-image",
      "family": "GPT Image 2.6",
      "description": "Editing built for the tightest control, edits scoped precisely to the instruction, with subject and composition preserved across many rounds of revision.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab1255d/vcqbyugh4tvlCOFkaWp1W.jpg",
      "price": "Text tokens (per 1M): $5.00 input, $1.25 cached, $10.00 output. Image tokens (per 1M): $8.00 input, $2.00 cached, $30.00 output. Changing the quality parameter significantly affects cost; by default we use high. Adjust it to your preference",
      "license": "commercial",
      "published": "2026-11-04",
      "url": "https://fal.ai/models/openai/gpt-image-2.6/sunburst/edit"
    },
    {
      "id": "openai/gpt-image-2.6/sunburst/text-to-image",
      "title": "GPT Image 2.6 Sunburst Text to Image",
      "category": "text-to-image",
      "family": "GPT Image 2.6",
      "description": "OpenAI's precision-focused image model, built for premium visual work, extra fidelity on intricate detail, in exchange for longer generation times.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab12553/DcOT-LMAkLJVxYuCGtSIX.jpg",
      "price": "Text tokens (per 1M): $5.00 input, $1.25 cached, $10.00 output. Image tokens (per 1M): $8.00 input, $2.00 cached, $30.00 output. Changing the quality parameter significantly affects cost; by default we use high. Adjust it to your preference",
      "license": "commercial",
      "published": "2026-11-04",
      "url": "https://fal.ai/models/openai/gpt-image-2.6/sunburst/text-to-image"
    },
    {
      "id": "elevenlabs/sound-effects/v3",
      "title": "ElevenLabs Sound Effects V3",
      "category": "text-to-audio",
      "description": "Generate sound effects up to 30 seconds from a text prompt with ElevenLabs SFX v3, with optional seamless looping.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab0e911/QbBs8ZoKkSHsA6_Xg62F-.jpg",
      "license": "commercial",
      "published": "2026-11-02",
      "url": "https://fal.ai/models/elevenlabs/sound-effects/v3"
    },
    {
      "id": "bytedance/seedance-2.5/us/draft/complete",
      "title": "Seedance 2.5 US Draft Complete",
      "category": "video-to-video",
      "family": "Seedance 2.5 US",
      "description": "Draft completion endpoint for Seedance 2.5 US - submit a draft id to regenerate the task at 1080p.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab079c2/LA6hgIjqVX_nrQ8SEkhlb.jpg",
      "price": "Your request will cost roughly $1.164 per second of generated video. Your request will cost roughly $0.0234 per 1,000 tokens for 1080p video. The number of tokens is roughly given by (height of output video * width of output video * (input ",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/bytedance/seedance-2.5/us/draft/complete"
    },
    {
      "id": "minimax/h3-max-turbo/reference-to-video",
      "title": "H3 Max Turbo Reference to Video",
      "category": "image-to-video",
      "family": "H3 Max",
      "description": "Generate video with synchronized audio from a prompt and image, video, or audio references with H3 Max Turbo, at 480p, 768p, or 1080p for up to 15 seconds.",
      "thumbnail": "https://v3b.fal.media/files/b/0aadb4ed/Nc_-PFaAPZ_9ikSRTCisF.jpg",
      "price": "Billing is calculated per second of output video, plus reference inputs beyond the included allowance. Video costs $0.05 per second at 480p, $0.08 per second at 768p, and $0.16 per second at 1080p. A 5-second 768p clip costs $0.40 before ad",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/minimax/h3-max-turbo/reference-to-video"
    },
    {
      "id": "mureka/api/generate/speech",
      "title": "Mureka Speech Generation",
      "category": "text-to-audio",
      "family": "mureka-api",
      "description": "Synthesizes speech from text using a built-in voice preset or an uploaded custom voice.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa61da/W8yYhfA_YR6yVzWvoIAbh.jpg",
      "price": "Billing is calculated per minute of generated audio, prorated to the exact output duration. Your request will cost $0.0817 per minute ($4.90 per hour), so a 30-second clip costs about $0.041. For $1 you can generate about 12 minutes of spee",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/api/generate/speech"
    },
    {
      "id": "mureka/api/song/edit",
      "title": "Mureka Song Editing",
      "category": "audio-to-audio",
      "family": "mureka-api",
      "description": "Rewrites a selected time region of a generated song or uploaded audio using replacement lyrics.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa61f3/3wLOBspCEcfdX8pDqVZA5.jpg",
      "price": "Billing is calculated per edit. Your request will cost $0.10 per edited region, and each request returns one edited song. For $1 you can run 10 edits.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/api/song/edit"
    },
    {
      "id": "mureka/api/song/stem",
      "title": "Mureka Stem Separation",
      "category": "audio-to-audio",
      "family": "mureka-api",
      "description": "Separates a song into downloadable audio stems, with available stem types and optional MIDI output determined by the selected separation model.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa6214/x2KU6ElS5zAlLVic8_wMj.jpg",
      "price": "Billing is calculated per separated song and depends on the selected model. With audio-separation-1, your request will cost $0.06 per song; with audio-separation-3 (the default), $0.20 per song; and with audio-separation-2, $0.70 per song.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/api/song/stem"
    },
    {
      "id": "mureka/api/voice/clone",
      "title": "Mureka Voice Cloning",
      "category": "audio-to-audio",
      "family": "Mureka 9.5",
      "description": "Creates a reusable custom voice from a short vocal recording and returns a voice identifier for supported generation workflows.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa6235/kH7JRPJYphhcNVbMoCM13.jpg",
      "price": "Billing is calculated per cloned voice. Your request will cost $5.00 per voice clone.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/api/voice/clone"
    },
    {
      "id": "mureka/v8/generate/instrumental",
      "title": "Mureka 8 Instrumental Generation",
      "category": "text-to-audio",
      "family": "Mureka 8",
      "description": "Generates instrumental music from a text description or uploaded instrumental reference, with support for multiple variations.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf6b91/Od9WiNXGMRlQgps9WYwQz.jpg",
      "price": "Billing is calculated per generated instrumental track. Your request will cost $0.045 per track, and each request returns one track. For $1 you can generate about 22 instrumental tracks.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v8/generate/instrumental"
    },
    {
      "id": "mureka/v8/generate/song",
      "title": "Mureka 8 Song Generation",
      "category": "text-to-audio",
      "family": "Mureka 8",
      "description": "Generates songs from a text prompt or supplied lyrics, with musical style, vocal, reference-audio, and melody controls, plus support for multiple variations.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf8dce/0lFfMSKWP3ocUh7XnGkO8.jpg",
      "price": "Billing is calculated per generated song. When you provide lyrics, your request will cost $0.045 per song; when you provide only a prompt without lyrics, it will cost $0.30 per song. Each request returns one song. For $1 you can generate ab",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v8/generate/song"
    },
    {
      "id": "mureka/v8/generate/soundtrack",
      "title": "Mureka 8 Soundtrack Generation",
      "category": "text-to-audio",
      "family": "Mureka 8",
      "description": "Generates soundtracks from an uploaded image or video, with optional style and mood guidance, audio-range selection, and multiple variations.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf8e6a/8Yz-UopodquIrDrrZ8K3z.jpg",
      "price": "Billing is calculated per generated soundtrack. Your request will cost $0.10 per soundtrack, and each request returns one track. For $1 you can generate 10 soundtracks.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v8/generate/soundtrack"
    },
    {
      "id": "mureka/v8/generate/track",
      "title": "Mureka 8 Track Generation",
      "category": "audio-to-audio",
      "family": "Mureka 8",
      "description": "Generates vocal or instrumental tracks for an existing song or uploaded audio, with controls for track type, musical style, generation region, and optional vocal lyrics.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf8eff/1tEZt1_oDHDlJAglrgzeU.jpg",
      "price": "Billing is calculated per generated track. Your request will cost $0.09 per track, regardless of the selected generation type. Each request returns one track. For $1 you can generate about 11 tracks.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v8/generate/track"
    },
    {
      "id": "mureka/v8/song/extend",
      "title": "Mureka 8 Song Extension",
      "category": "audio-to-audio",
      "family": "Mureka 8",
      "description": "Extends a generated song or uploaded audio from a specified timestamp using supplied lyrics, with head or tail extension depending on the selected model.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf8fb4/v6gy1QvfavujHKZWT4w8k.jpg",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v8/song/extend"
    },
    {
      "id": "mureka/v8/song/remix",
      "title": "Mureka 8 Song Remix",
      "category": "audio-to-audio",
      "family": "Mureka 8",
      "description": "Remixes a generated song or uploaded audio with Mureka 8, using complete lyrics and a prompt that controls genre, mood, instrumentation, and arrangement. Accepts source audio from 10 to 350 seconds; vocal-only, accompaniment-only, and instrumental tracks are not supported.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa620b/dzA6SjokYI-4v7xeUE9wD.jpg",
      "price": "Billing is calculated per remixed song. Your request will cost $0.20 per song, and each request returns one song. For $1 you can remix about 5 songs.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v8/song/remix"
    },
    {
      "id": "mureka/v9.5/generate/instrumental",
      "title": "Mureka 9.5 Instrumental Generation",
      "category": "text-to-audio",
      "family": "Mureka 9.5",
      "description": "Generates instrumental music from a text description or uploaded instrumental reference, with support for multiple variations.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa6193/e-XLyqpAgL2AXHiNGr1Fz.jpg",
      "price": "Billing is calculated per generated instrumental track. Your request will cost $0.15 per track, and each request returns one track. For $1 you can generate about 6 instrumental tracks.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v9.5/generate/instrumental"
    },
    {
      "id": "mureka/v9.5/generate/song",
      "title": "Mureka 9.5 Song Generation",
      "category": "text-to-audio",
      "family": "Mureka 9.5",
      "description": "Generates songs from a text prompt or supplied lyrics, with musical style, vocal, reference-audio, and melody controls, plus support for multiple variations.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa61bd/nUKHfwGHS0eFH_BoR-mDL.jpg",
      "price": "Billing is calculated per generated song. When you provide lyrics, your request will cost $0.15 per song; when you provide only a prompt without lyrics, it will cost $0.50 per song. Each request returns one song. For $1 you can generate abo",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v9.5/generate/song"
    },
    {
      "id": "mureka/v9.5/song/remix",
      "title": "Mureka 9.5 Song Remix",
      "category": "audio-to-audio",
      "family": "Mureka 9.5",
      "description": "Remixes a generated song or uploaded audio with Mureka 9.5, using complete lyrics and a prompt that controls genre, mood, instrumentation, arrangement, and vocal gender. Accepts source audio from 10 to 480 seconds, including vocal-only, accompaniment-only, and instrumental tracks.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab0fcc0/VS1gBrnLqCVGJ3O3xadjJ_lzkAXzaN.png",
      "price": "Billing is calculated per remixed song. Your request will cost $0.50 per song, and each request returns one song. For $1 you can remix about 2 songs.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v9.5/song/remix"
    },
    {
      "id": "mureka/v9/generate/instrumental",
      "title": "Mureka 9 Instrumental Generation",
      "category": "text-to-audio",
      "family": "Mureka 9",
      "description": "Generates instrumental music from a text description or uploaded instrumental reference, with support for multiple variations.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaff1b0/QdgRyEDZMctEvdrsBhiPQ.jpg",
      "price": "Billing is calculated per generated instrumental track. Your request will cost $0.045 per track, and each request returns one track. For $1 you can generate about 22 instrumental tracks.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v9/generate/instrumental"
    },
    {
      "id": "mureka/v9/generate/song",
      "title": "Mureka 9 Song Generation",
      "category": "text-to-audio",
      "family": "Mureka 9",
      "description": "Generates songs from a text prompt or supplied lyrics, with musical style, vocal, reference-audio, and melody controls, plus support for multiple variations.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaff182/4drxC0q8eOOC6uSU4RyCQ.jpg",
      "price": "Billing is calculated per generated song. When you provide lyrics, your request will cost $0.045 per song; when you provide only a prompt without lyrics, it will cost $0.30 per song. Each request returns one song. For $1 you can generate ab",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v9/generate/song"
    },
    {
      "id": "mureka/v9/generate/soundtrack",
      "title": "Mureka 9 Soundtrack Generation",
      "category": "audio-to-audio",
      "family": "Mureka 9",
      "description": "Generates soundtracks from an uploaded image or video, with optional style and mood guidance, audio-range selection, and multiple variations.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaff1ef/_Rj7C3h4nbFzIekBD9VY6.jpg",
      "price": "Billing is calculated per generated soundtrack. Your request will cost $0.10 per soundtrack, whether it is guided by an image or a video. Each request returns one soundtrack. For $1 you can generate 10 soundtracks.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/mureka/v9/generate/soundtrack"
    },
    {
      "id": "speridlabs/idol",
      "title": "Sperid Labs Idol",
      "category": "image-to-3d",
      "family": "Idol",
      "description": "Idol by Sperid Labs turns a single image of an object into a 3D Gaussian splat, with a GLB point-cloud preview.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab07f87/Z7OKjkqz5tdOAEH34KUOn.jpg",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/speridlabs/idol"
    },
    {
      "id": "tbc/mischief-1/image-to-video",
      "title": "Mischief 1",
      "category": "image-to-video",
      "description": "TBC mischief 1 is the world's first neuron-derived video model. Generate videos 5x faster at a fraction of the cost, so you can explore more ideas, experiment freely, and make more mischief.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab02f0b/jGztSDlBc8z00gEEY1COZ.jpg",
      "price": "Billing is calculated per second of video generated. For 480p, the cost is $0.005 per second; for 720p, $0.01 per second. A 5-second video costs $0.025 at 480p or $0.05 at 720p.",
      "license": "commercial",
      "published": "2026-10-30",
      "url": "https://fal.ai/models/tbc/mischief-1/image-to-video"
    },
    {
      "id": "fal-ai/motion-to-code/glsl",
      "title": "Motion To Code",
      "category": "video-to-video",
      "family": "motion-to-code",
      "description": "Turn a prompt, an image or a reference video into a seamlessly looping GLSL shader, with a page that plays it live.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab01c04/-wDxOaNu4wFkcIbseaQTv.jpg",
      "price": "Your request will cost $0.30 per second of animation when MiniMax H3 Max animates your prompt or image first, or $0.20 per second from your own video, with a 5-second minimum: a 6-second loop from a prompt costs $1.80, a 10-second clip rebu",
      "license": "commercial",
      "published": "2026-10-29",
      "url": "https://fal.ai/models/fal-ai/motion-to-code/glsl"
    },
    {
      "id": "fal-ai/motion-to-code/gsap",
      "title": "Motion To Code",
      "category": "video-to-video",
      "family": "motion-to-code",
      "description": "Turn a prompt, an image or a reference video into an editable GSAP animation: one self-contained HTML page you can play, tweak and ship.",
      "thumbnail": "https://v3b.fal.media/files/b/0ab01c1a/wBfWUwY5LJJJQJ5AfdIns.jpg",
      "price": "Your request will cost $0.30 per second of animation when MiniMax H3 Max animates your prompt or image first, or $0.20 per second from your own video, with a 5-second minimum: a 6-second loop from a prompt costs $1.80, a 10-second clip rebu",
      "license": "commercial",
      "published": "2026-10-29",
      "url": "https://fal.ai/models/fal-ai/motion-to-code/gsap"
    },
    {
      "id": "blackforestlabs/flux-3/klein/base/edit-image/lora",
      "title": "FLUX 3 Klein Base Edit LoRA",
      "category": "image-to-image",
      "family": "Flux 3",
      "description": "Edit images with FLUX 3 Image Klein base and up to 3 custom LoRAs.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaff6e7/s0s9e0rtCaEhT1ocPQr9k.jpg",
      "price": "Your request will cost $0.035 per megapixel of output. A 1024x1024 image costs $0.035.",
      "license": "commercial",
      "published": "2026-10-26",
      "url": "https://fal.ai/models/blackforestlabs/flux-3/klein/base/edit-image/lora"
    },
    {
      "id": "blackforestlabs/flux-3/klein/base/text-to-image/lora",
      "title": "FLUX 3 Klein Base Text to Image LoRA",
      "category": "text-to-image",
      "family": "Flux 3",
      "description": "Generate images with FLUX 3 Image Klein base and up to 3 custom LoRAs.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaff6da/8YEUun7ENziHr_ZRPlZw2.jpg",
      "price": "Your request will cost $0.035 per megapixel of output. A 1024x1024 image costs $0.035.",
      "license": "commercial",
      "published": "2026-10-26",
      "url": "https://fal.ai/models/blackforestlabs/flux-3/klein/base/text-to-image/lora"
    },
    {
      "id": "blackforestlabs/flux-3/klein/edit-image/lora",
      "title": "FLUX 3 Klein Edit LoRA",
      "category": "image-to-image",
      "family": "Flux 3",
      "description": "Edit images fast with distilled FLUX 3 Image Klein and up to 3 custom LoRAs.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaff72d/bsuk_ypo69hQhRFgl6eQw.jpg",
      "price": "Your request will cost $0.012 per megapixel of output. A 1024x1024 image costs $0.012.",
      "license": "commercial",
      "published": "2026-10-26",
      "url": "https://fal.ai/models/blackforestlabs/flux-3/klein/edit-image/lora"
    },
    {
      "id": "blackforestlabs/flux-3/klein/text-to-image/lora",
      "title": "FLUX 3 Klein Text to Image LoRA",
      "category": "text-to-image",
      "family": "Flux 3",
      "description": "Generate images fast with distilled FLUX 3 Image Klein and up to 3 custom LoRAs.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaff718/6qt7rn9vNChTVNnFaFMcQ.jpg",
      "price": "Your request will cost $0.012 per megapixel of output. A 1024x1024 image costs $0.012.",
      "license": "commercial",
      "published": "2026-10-26",
      "url": "https://fal.ai/models/blackforestlabs/flux-3/klein/text-to-image/lora"
    },
    {
      "id": "ideogram/vector/image-to-vector",
      "title": "Ideogram Vector",
      "category": "image-to-image",
      "family": "ideogram-vector",
      "description": "Converts images into editable SVG vector graphics, with a default mode for finer details and a simple mode for fewer shapes.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf6b9a/rhJnz9ybiIoj5nf9vY178.jpg",
      "price": "Billing is $0.10 per generated SVG, in both default and simple modes. Converting one image into one SVG costs $0.10.",
      "license": "commercial",
      "published": "2026-10-22",
      "url": "https://fal.ai/models/ideogram/vector/image-to-vector"
    },
    {
      "id": "ideogram/vector/text-to-vector",
      "title": "Ideogram Vector",
      "category": "text-to-image",
      "family": "ideogram-vector",
      "description": "Generates editable SVG vector graphics from text prompts, with optional reference images, three quality levels, and aspect ratio controls.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf6b95/3R80BKx20coG5pZ1RPHPO.jpg",
      "price": "Billing is per generated SVG: $0.13 for low, $0.16 for medium, and $0.20 for high quality. With reference images, high quality costs $0.32 per SVG; low and medium prices are unchanged. For example, 2 medium-quality SVGs cost $0.32.",
      "license": "commercial",
      "published": "2026-10-22",
      "url": "https://fal.ai/models/ideogram/vector/text-to-vector"
    },
    {
      "id": "heygen/heygen-video/image-to-video",
      "title": "HeyGen Video Image to Video",
      "category": "image-to-video",
      "family": "HeyGen Video",
      "description": "HeyGen Video animates your image as the first frame of a 5 to 15 second clip, with generated sound, built on MiniMax H3 and post-trained by HeyGen.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf314c/yeLGH-MM-_N78NxqoZ9Rj.jpg",
      "price": "Limited-time promo (50% off) until October 31, 2026: video costs $0.01 per second at 480p and 768p, and $0.03 per second at 1080p and 2K (regular price: $0.02 and $0.06 per second).",
      "license": "commercial",
      "published": "2026-10-21",
      "url": "https://fal.ai/models/heygen/heygen-video/image-to-video"
    },
    {
      "id": "heygen/heygen-video/reference-to-video",
      "title": "HeyGen Video Reference to Video",
      "category": "image-to-video",
      "family": "HeyGen Video",
      "description": "HeyGen Video builds a new 5 to 15 second scene with sound that keeps the products, people or places from your reference images, videos or audio.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf3125/dgnBBX3mq3uzXTqTgOuOM.jpg",
      "price": "Limited-time promo (50% off) until October 31, 2026: video costs $0.01 per second at 480p and 768p, and $0.03 per second at 1080p and 2K (regular price: $0.02 and $0.06 per second).",
      "license": "commercial",
      "published": "2026-10-21",
      "url": "https://fal.ai/models/heygen/heygen-video/reference-to-video"
    },
    {
      "id": "heygen/heygen-video/text-to-video",
      "title": "HeyGen Video Text to Video",
      "category": "text-to-video",
      "family": "HeyGen Video",
      "description": "HeyGen Video, built on MiniMax H3 and post-trained by HeyGen, turns a prompt into a 5 to 15 second clip with its own sound.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf310e/ILb3IF4i4sFXwWa_JvJca.jpg",
      "price": "Limited-time promo (50% off) until October 31, 2026: video costs $0.01 per second at 480p and 768p, and $0.03 per second at 1080p and 2K (regular price: $0.02 and $0.06 per second).",
      "license": "commercial",
      "published": "2026-10-21",
      "url": "https://fal.ai/models/heygen/heygen-video/text-to-video"
    },
    {
      "id": "heygen/heygen-voice",
      "title": "HeyGen Voice",
      "category": "text-to-speech",
      "family": "HeyGen Voice",
      "description": "HeyGen Voice turns text into natural speech with a library voice or an instant clone from a reference clip.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf3121/SYQYxjbCy5qaMJKYlhY8y.jpg",
      "price": "Limited-time promo (50% off) until October 31, 2026: speech costs $0.015 per 1,000 characters of input text (regular price: $0.03).",
      "license": "commercial",
      "published": "2026-10-21",
      "url": "https://fal.ai/models/heygen/heygen-voice"
    },
    {
      "id": "heygen/heygen-voice/turbo",
      "title": "HeyGen Voice Turbo",
      "category": "text-to-speech",
      "family": "HeyGen Voice",
      "description": "HeyGen Voice Turbo is the fast, low-cost tier of HeyGen Voice, with library voices and instant cloning from a reference clip.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf3122/Zv-3HScnTaRrOrw9g3ze-.jpg",
      "price": "Limited-time promo (50% off) until October 31, 2026: speech costs $0.002 per 1,000 characters of input text (regular price: $0.004).",
      "license": "commercial",
      "published": "2026-10-21",
      "url": "https://fal.ai/models/heygen/heygen-voice/turbo"
    },
    {
      "id": "kling/v4/image-to-video",
      "title": "Kling 4.0 Image to Video",
      "category": "image-to-video",
      "family": "Kling v4",
      "description": "Kling 4.0 animates a starting image into a 3 to 30-second video at 720p or 1080p, with optional ending frames and per-shot prompts.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaea00f/kDSSuRNelHs4Q40pk4QFR.jpg",
      "price": "Video costs $0.157 per second at 720p and $0.235 per second at 1080p.",
      "license": "commercial",
      "published": "2026-10-21",
      "url": "https://fal.ai/models/kling/v4/image-to-video"
    },
    {
      "id": "kling/v4/text-to-video",
      "title": "Kling 4.0 Text to Video",
      "category": "text-to-video",
      "family": "Kling v4",
      "description": "Kling 4.0 generates videos from text prompts at 720p or 1080p, with durations from 3 to 30 seconds and support for per-shot prompts.",
      "thumbnail": "https://v3b.fal.media/files/b/0aae9ff4/BBn7FmeY6HUb0EJgkXkWJ.jpg",
      "price": "Video costs $0.157 per second at 720p and $0.235 per second at 1080p.",
      "license": "commercial",
      "published": "2026-10-21",
      "url": "https://fal.ai/models/kling/v4/text-to-video"
    },
    {
      "id": "bria/fibo-2/edit",
      "title": "Fibo 2.0 Edit",
      "category": "image-to-image",
      "family": "Fibo",
      "description": "Fibo 2.0 Edit makes targeted, instruction-based edits from up to four images. Background swaps, product restaging, keeping the product true to the original. Trained only on licensed data, with clear commercial rights.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf018f/r_IIxW5iTEj8Z1mQQQesQ.jpg",
      "price": "Billing is calculated per edited image. Your request will cost $0.04 per image. For example, editing 10 images costs $0.40.",
      "license": "commercial",
      "published": "2026-10-20",
      "url": "https://fal.ai/models/bria/fibo-2/edit"
    },
    {
      "id": "bria/fibo-2/generate",
      "title": "Fibo 2.0 Generate",
      "category": "text-to-image",
      "family": "Fibo",
      "description": "Fibo 2.0 Generate is Bria's text-to-image model, strongest on reasoning and knowledge-heavy prompts the ones that need real understanding, not just a style match. Trained only on licensed data, with clear commercial rights.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf018b/OBERerkifKAq4LsOG1Eed.jpg",
      "price": "Billing is calculated per generated image. Each approximately 1MP image costs $0.04. For example, generating 10 images costs $0.40.",
      "license": "commercial",
      "published": "2026-10-20",
      "url": "https://fal.ai/models/bria/fibo-2/generate"
    },
    {
      "id": "blackforestlabs/flux-3-max/image-to-video",
      "title": "FLUX 3 Max Image to Video",
      "category": "image-to-video",
      "family": "Flux 3 Max",
      "description": "fal’s FLUX 3 Max is a post-trained variant of FLUX 3 Video, tuned for stronger prompt adherence and better visual quality while optimized for lower latency.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaf0ae2/vqNYPsX9h49SQulyxqJyL.jpg",
      "price": "Billing is calculated per second of video generated. Through November 7, 2026 (UTC), the 75% discount brings the rate to $0.0425 per second at 720p and $0.0725 per second at 1080p. From November 8 at 00:00 UTC through November 22, 2026, the",
      "license": "commercial",
      "published": "2026-10-19",
      "url": "https://fal.ai/models/blackforestlabs/flux-3-max/image-to-video"
    },
    {
      "id": "blackforestlabs/flux-3-max/keyframes-to-video",
      "title": "FLUX 3 Max Keyframes to Video",
      "category": "image-to-video",
      "family": "Flux 3 Max",
      "description": "fal’s FLUX 3 Max is a post-trained variant of FLUX 3 Video, tuned for stronger prompt adherence and better visual quality while optimized for lower latency.",
      "thumbnail": "https://v3b.fal.media/files/b/0aae8cf6/-Cuw8bwhTM9rdnwq7eICJ.jpg",
      "price": "Billing is calculated per second of video generated. From November 8, 2026 at 00:00 UTC through November 22, 2026, the 50% promotion brings the rate to $0.085 per second at 720p and $0.145 per second at 1080p. During this promotion, a 10-se",
      "license": "commercial",
      "published": "2026-10-19",
      "url": "https://fal.ai/models/blackforestlabs/flux-3-max/keyframes-to-video"
    },
    {
      "id": "blackforestlabs/flux-3-max/text-to-video",
      "title": "FLUX 3 Max Text to Video",
      "category": "text-to-video",
      "family": "Flux 3 Max",
      "description": "fal’s FLUX 3 Max is a post-trained variant of FLUX 3 Video, tuned for stronger prompt adherence and better visual quality while optimized for lower latency.",
      "thumbnail": "https://v3b.fal.media/files/b/0aae8ccd/cfm_KIh1OltujxYuHIzzV.jpg",
      "price": "Billing is calculated per second of video generated. Through November 7, 2026 (UTC), the 75% discount brings the rate to $0.0425 per second at 720p and $0.0725 per second at 1080p. From November 8 at 00:00 UTC through November 22, 2026, the",
      "license": "commercial",
      "published": "2026-10-19",
      "url": "https://fal.ai/models/blackforestlabs/flux-3-max/text-to-video"
    },
    {
      "id": "fal-ai/moss-image",
      "title": "MOSS Image",
      "category": "text-to-image",
      "family": "moss-image",
      "description": "MOSS-Image-SFT is a text-to-image model specializing in Chinese and English text rendering and structured layout control for precise text placement and object arrangement.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaee64a/UHOuCdOU1lfOov57OOYN-.jpg",
      "price": "Your request will cost $0.05 per megapixel",
      "license": "commercial",
      "published": "2026-10-18",
      "url": "https://fal.ai/models/fal-ai/moss-image"
    },
    {
      "id": "blackforestlabs/flux-3/klein/edit-image",
      "title": "FLUX.3 Klein Image Editing",
      "category": "image-to-image",
      "family": "Flux 3",
      "description": "Edits and composes images from text instructions and up to ten references with FLUX.3 Klein, with optional bounding-box controls for object placement and transformation.",
      "thumbnail": "https://v3b.fal.media/files/b/0aae6141/RHrU_CVUIV0D1xwgnu0Dk.jpg",
      "price": "Pricing per output image is $0.008 at 768 × 768, $0.009 at 1K, $0.011 at 1.5K, $0.014 at 2K, and $0.036 at 4K. The same planned rates apply to generation and editing; for example, 10 images at 1K would cost $0.09.",
      "license": "commercial",
      "published": "2026-10-15",
      "url": "https://fal.ai/models/blackforestlabs/flux-3/klein/edit-image"
    },
    {
      "id": "blackforestlabs/flux-3/klein/text-to-image",
      "title": "FLUX.3 Klein",
      "category": "text-to-image",
      "family": "Flux 3",
      "description": "Generates images from text prompts with FLUX.3 Klein, with configurable image size, reproducible seeds, prompt expansion and optional bounding-box layout control.",
      "thumbnail": "https://v3b.fal.media/files/b/0aae6139/Z-1wSL9wkWSud1hxPjTL4.jpg",
      "price": "Pricing per output image is $0.008 at 768 × 768, $0.009 at 1K, $0.011 at 1.5K, $0.014 at 2K, and $0.036 at 4K. The same planned rates apply to generation and editing; for example, 10 images at 1K would cost $0.09.",
      "license": "commercial",
      "published": "2026-10-15",
      "url": "https://fal.ai/models/blackforestlabs/flux-3/klein/text-to-image"
    },
    {
      "id": "bytedance/omnihuman/v2",
      "title": "Bytedance Omnihuman V2",
      "category": "image-to-video",
      "family": "Omnihuman 2",
      "description": "Omnihuman v2 generates video using an image of a human figure paired with an audio file. It produces vivid, high-quality videos where the character’s emotions and movements maintain a strong correlation with the audio.",
      "thumbnail": "https://v3b.fal.media/files/b/0aae7dce/7Dya2scYo84ixOeq8iqRt.jpg",
      "license": "commercial",
      "published": "2026-10-15",
      "url": "https://fal.ai/models/bytedance/omnihuman/v2"
    },
    {
      "id": "kotoba-technologies/kotoba-streaming-stt/v1/realtime",
      "title": "Kotoba Realtime Speech-to-Text",
      "category": "speech-to-text",
      "family": "Kotoba Streaming STT",
      "description": "Live transcription over WebSocket, with partial results as the speaker talks.",
      "thumbnail": "https://v3b.fal.media/files/b/0aadb23d/LgCh9ClUBK_Ld6ubdxrbz.jpg",
      "price": "Kotoba Streaming STT: $0.39 per hour of input audio for the one-shot demo; billed per input audio second.",
      "license": "commercial",
      "published": "2026-10-15",
      "url": "https://fal.ai/models/kotoba-technologies/kotoba-streaming-stt/v1/realtime"
    },
    {
      "id": "kotoba-technologies/kotoba-stt/v1/speech-to-text",
      "title": "Kotoba Speech-to-Text",
      "category": "speech-to-text",
      "family": "Kotoba STT",
      "description": "Accurate transcription of an audio clip in one request, with word timestamps, tuned for Japanese and East Asian languages.",
      "thumbnail": "https://v3b.fal.media/files/b/0aadb23c/t5NyD12xHgFCxg6SaeNFZ.jpg",
      "price": "Kotoba STT: $0.25 per hour of input audio",
      "license": "commercial",
      "published": "2026-10-15",
      "url": "https://fal.ai/models/kotoba-technologies/kotoba-stt/v1/speech-to-text"
    },
    {
      "id": "kotoba-technologies/kotoba-tts/v2/tts/ws",
      "title": "Kotoba Text-to-Speech",
      "category": "text-to-speech",
      "family": "Kotoba TTS",
      "description": "Natural Japanese speech, streamed back as it is generated so playback starts almost immediately.",
      "thumbnail": "https://v3b.fal.media/files/b/0aadb23d/pIOoy7DcpoEZ8Qfm7d0Nh.jpg",
      "price": "Kotoba TTS: $0.015 per 1,000 UTF-8 bytes of input text; 1 billable unit = 1,000 UTF-8 bytes.",
      "license": "commercial",
      "published": "2026-10-15",
      "url": "https://fal.ai/models/kotoba-technologies/kotoba-tts/v2/tts/ws"
    },
    {
      "id": "krea/v3/edit",
      "title": "Krea 3 Edit",
      "category": "image-to-image",
      "family": "Krea 3",
      "description": "Edits and combines reference images using text instructions, with optional prompt expansion, output-size controls, and reproducible generation through a seed.",
      "thumbnail": "https://v3b.fal.media/files/b/0aae586e/Yv5jrU5Sim8QifUup4Ah2.jpg",
      "price": "Billed per megapixel at $0.025/MP, counting the output image plus each reference image as the model processes it (references larger than the output are counted at the output’s size). 1 MP = 1024×1024 pixels.",
      "license": "commercial",
      "published": "2026-10-15",
      "url": "https://fal.ai/models/krea/v3/edit"
    },
    {
      "id": "krea/v3/realtime",
      "title": "Krea 3 Realtime",
      "category": "image-to-image",
      "family": "Krea 3",
      "description": "Edits an input image in a realtime session using text prompts, with seed control.",
      "thumbnail": "https://v3b.fal.media/files/b/0aae5886/cBCX0ma5XJS6HvjLXo1np.jpg",
      "price": "Current pricing is $0.013 per second of a realtime session; for example, 60 seconds cost $0.78. Final Krea 3 pricing has not been set and may change.",
      "license": "commercial",
      "published": "2026-10-15",
      "url": "https://fal.ai/models/krea/v3/realtime"
    },
    {
      "id": "krea/v3/text-to-image",
      "title": "Krea 3 Text to Image",
      "category": "text-to-image",
      "family": "Krea 3",
      "description": "Generates images from text prompts with optional prompt expansion, configurable output dimensions, and seed control.",
      "thumbnail": "https://v3b.fal.media/files/b/0aae5877/x3ZWk-bA0lnIAGHb5ligG.jpg",
      "price": "Billed per megapixel at $0.025/MP, counting the output image plus each reference image as the model processes it (references larger than the output are counted at the output’s size). 1 MP = 1024×1024 pixels.",
      "license": "commercial",
      "published": "2026-10-15",
      "url": "https://fal.ai/models/krea/v3/text-to-image"
    },
    {
      "id": "mirelo-ai/sfx2.0/text-to-audio",
      "title": "Mirelo SFX2.0",
      "category": "text-to-audio",
      "family": "SFX 2.0",
      "description": "Generates sound effects and ambience from a text description, up to 600 seconds long, with up to 4 variations per request.",
      "thumbnail": "https://v3b.fal.media/files/b/0aae5475/0QvHHPHbJyAhC7JkgQDHq.jpg",
      "price": "$0.010 per second through December 13, 2026 (33% off the standard $0.015 per second), then $0.015 per second. Billed per second of generated audio, per sample.",
      "license": "commercial",
      "published": "2026-10-14",
      "url": "https://fal.ai/models/mirelo-ai/sfx2.0/text-to-audio"
    },
    {
      "id": "mureka/api/generate/podcast",
      "title": "Mureka Podcast Generation",
      "category": "text-to-audio",
      "family": "mureka-api",
      "description": "Generates podcast audio from an ordered sequence of conversation turns, with text and a voice selection for each turn.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa61ce/rxfwDlsIueQFyAUppFLNV.jpg",
      "price": "Billing is calculated per minute of generated audio, prorated to the exact output duration. Your request will cost $0.115 per minute ($6.90 per hour), so a 10-minute podcast costs $1.15. For $1 you can generate about 8.7 minutes of podcast ",
      "license": "commercial",
      "published": "2026-10-07",
      "url": "https://fal.ai/models/mureka/api/generate/podcast"
    },
    {
      "id": "fal-ai/kandinsky6-lite/text-to-video",
      "title": "Kandinsky 6.0 Lite",
      "category": "text-to-video",
      "family": "Kandinsky 6.0 Lite",
      "description": "Kandinsky 6.0 Lite is a lightweight, high-speed text-to-video model from Kandinsky Lab, built for fast, efficient generation with strong prompt adherence.",
      "thumbnail": "https://v3b.fal.media/files/b/0aad28fa/A2SiDGpqWJmGTm_X2yuqg.jpg",
      "price": "A 5-second 480p video costs $0.16 at the default 10 steps. 4 steps cost $0.08, 16 steps cost $0.24, and each extra step adds about $0.013. Upscaling (16:9) adds $0.12 for 2×, $0.16 for 1080p and $0.49 for 4×, so at 10 steps these cost $0.28",
      "license": "commercial",
      "published": "2026-10-06",
      "url": "https://fal.ai/models/fal-ai/kandinsky6-lite/text-to-video"
    },
    {
      "id": "fal-ai/kandinsky6-pro/text-to-video",
      "title": "Kandinsky 6.0 Pro",
      "category": "text-to-video",
      "family": "Kandinsky 6.0 Pro",
      "description": "Kandinsky 6.0 Pro is Kandinsky Lab's flagship text-to-video model, generating high-resolution clips with cinematic motion and precise prompt adherence.",
      "thumbnail": "https://v3b.fal.media/files/b/0aad2929/Z9eM0SNw8Gulk3vXDU5K8.jpg",
      "price": "A 5-second 480p video costs $1.35 at the default 50 steps. 20 steps cost $0.56, 8 steps cost $0.24, and each step adds about $0.027. Upscaling (16:9) adds $0.14 for 2×, $0.18 for 1080p and $0.56 for 4×, so at 50 steps these cost $1.49, $1.5",
      "license": "commercial",
      "published": "2026-10-06",
      "url": "https://fal.ai/models/fal-ai/kandinsky6-pro/text-to-video"
    },
    {
      "id": "google/nano-banana-2.1",
      "title": "Nano Banana 2.1",
      "category": "text-to-image",
      "family": "Nano Banana 2.1",
      "description": "Nano Banana 2.1 by Google generates images from text prompts, with output resolutions up to 4K, adjustable aspect ratios, and optional web search grounding.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaad149/0HLWjH4WLjtFUmi3bYfF4.jpg",
      "price": "Text/thinking tokens (per 1M): $1.764 input, $8.823 output. Image tokens (per 1M): $1.764 input, $35.294 output. Changing the resolution parameter significantly affects cost. See the description at the bottom of this page for more details o",
      "license": "commercial",
      "published": "2026-10-06",
      "url": "https://fal.ai/models/google/nano-banana-2.1"
    },
    {
      "id": "xai/grok-imagine-video/v1.5/lite/text-to-video",
      "title": "Grok Imagine Video 1.5 Lite Text to Video",
      "category": "text-to-video",
      "family": "Grok Imagine 1.5 Lite",
      "description": "Generate videos from text prompts using xAI's Grok Imagine Video 1.5 Lite model.",
      "thumbnail": "https://v3b.fal.media/files/b/0aac8c97/q0dnsNb5Wra4NzkWywH-S.jpg",
      "price": "Priced per second of output video, by resolution: 480p at $0.02/sec, 720p at $0.03/sec, 1080p at $0.14/sec. A 5-second 480p clip costs $0.10; 720p costs $0.15; 1080p costs $0.70.",
      "license": "commercial",
      "published": "2026-10-01",
      "url": "https://fal.ai/models/xai/grok-imagine-video/v1.5/lite/text-to-video"
    },
    {
      "id": "veed/clean-audio",
      "title": "VEED Clean Audio",
      "category": "audio-to-audio",
      "description": "Studio-quality speech from noisy recordings",
      "thumbnail": "https://v3b.fal.media/files/b/0aab6e5b/ZKvDjaHYYnJDJrJz4WyOu.jpg",
      "price": "Billing is calculated at $0.0125 per minute of input audio. Duration is rounded to whole seconds, then up to full minutes, with a 1-minute minimum. For example, 10 minutes costs $0.125, and 30 minutes costs $0.375.",
      "license": "commercial",
      "published": "2026-09-30",
      "url": "https://fal.ai/models/veed/clean-audio"
    },
    {
      "id": "elevenlabs/tts/eleven-v4",
      "title": "Eleven v4",
      "category": "text-to-speech",
      "family": "Eleven v4",
      "description": "Generate expressive speech with Eleven v4 from ElevenLabs. Control delivery with audio tags, voice stability, similarity settings, and IPA pronunciation.",
      "thumbnail": "https://v3b.fal.media/files/b/0aac4241/SUZvoStueCW3Z6uEtzMiA.jpg",
      "license": "commercial",
      "published": "2026-09-28",
      "url": "https://fal.ai/models/elevenlabs/tts/eleven-v4"
    },
    {
      "id": "elevenlabs/tts/eleven-v4-turbo",
      "title": "Eleven v4 Turbo",
      "category": "text-to-speech",
      "family": "Eleven v4",
      "description": "Generate speech with Eleven v4 Turbo from ElevenLabs. Choose a voice and control delivery with audio tags, stability, similarity settings, and IPA pronunciation.",
      "thumbnail": "https://v3b.fal.media/files/b/0aac4243/7KzAVHqtn2QWjIY5QWFG_.jpg",
      "license": "commercial",
      "published": "2026-09-28",
      "url": "https://fal.ai/models/elevenlabs/tts/eleven-v4-turbo"
    },
    {
      "id": "google/gemini-3.8-flash-lite-tts",
      "title": "Gemini 3.8 Flash Lite TTS",
      "category": "text-to-speech",
      "family": "TTS",
      "description": "Generate expressive speech with Gemini 3.8 Flash Lite TTS. Choose from 30 voices, guide delivery with style instructions, and create single-speaker narration or two-speaker dialogue.",
      "thumbnail": "https://v3b.fal.media/files/b/0aabd888/nA5lFvHKT-YnZx9tkjubZ.jpg",
      "license": "commercial",
      "published": "2026-09-25",
      "url": "https://fal.ai/models/google/gemini-3.8-flash-lite-tts"
    },
    {
      "id": "google/gemini-3.8-flash-tts",
      "title": "Gemini 3.8 Flash TTS",
      "category": "text-to-speech",
      "family": "TTS",
      "description": "Generate expressive speech with Gemini 3.8 Flash TTS. Choose from 30 voices, direct delivery with style instructions, and create single-speaker narration or two-speaker dialogue.",
      "thumbnail": "https://v3b.fal.media/files/b/0aabd875/3gAgS2GOPVAkwRanZZpN7.jpg",
      "license": "commercial",
      "published": "2026-09-25",
      "url": "https://fal.ai/models/google/gemini-3.8-flash-tts"
    },
    {
      "id": "meshy/v7.1/image-to-3d",
      "title": "Meshy 7.1 Image to 3D",
      "category": "image-to-3d",
      "family": "V7.1",
      "description": "Meshy 7.1 is the latest standard model from Meshy. It generates 3D models from a single image, with optional textures and PBR maps.",
      "thumbnail": "https://v3b.fal.media/files/b/0aab191a/G-wnzTid6Od_ewQ3ASuOF.jpg",
      "price": "A base model without textures costs $0.80. Adding textures brings the total to $1.20. Optional auto-rigging adds $0.20, and animation adds $0.12 per call. A textured model with auto-rigging and animation costs $1.52.",
      "license": "commercial",
      "published": "2026-09-21",
      "url": "https://fal.ai/models/meshy/v7.1/image-to-3d"
    },
    {
      "id": "meshy/v7.1/multi-image-to-3d",
      "title": "Meshy 7.1 Multi Image to 3D",
      "category": "image-to-3d",
      "family": "V7.1",
      "description": "Meshy 7.1 is the latest standard model from Meshy. It generates 3D models from one to four views of the same object, with optional textures and PBR maps.",
      "thumbnail": "https://v3b.fal.media/files/b/0aab191e/ASgJ3AifwlPiIKDKJRBd1.jpg",
      "price": "A base model without textures costs $0.80. Adding textures brings the total to $1.20. Optional auto-rigging adds $0.20, and animation adds $0.12 per call. A textured model with auto-rigging and animation costs $1.52.",
      "license": "commercial",
      "published": "2026-09-21",
      "url": "https://fal.ai/models/meshy/v7.1/multi-image-to-3d"
    },
    {
      "id": "meshy/v7.1/text-to-3d",
      "title": "Meshy 7.1 Text to 3D",
      "category": "text-to-3d",
      "family": "V7.1",
      "description": "Meshy 7.1 is the latest standard model from Meshy. It generates 3D models from text prompts, with untextured preview and textured full modes.",
      "thumbnail": "https://v3b.fal.media/files/b/0aab15a0/--xBfX2nTr0EhQxGNw2Bw.jpg",
      "price": "A base model without textures costs $0.80. Adding textures brings the total to $1.20. Optional auto-rigging adds $0.20, and animation adds $0.12 per call. A textured model with auto-rigging and animation costs $1.52.",
      "license": "commercial",
      "published": "2026-09-21",
      "url": "https://fal.ai/models/meshy/v7.1/text-to-3d"
    },
    {
      "id": "minimax/h3-max/styles/hand-drawn",
      "title": "H3 Max Hand Drawn",
      "category": "text-to-video",
      "family": "H3 Max",
      "description": "Generates 768p video with audio in a hand-drawn animation style from text prompts or an optional first-frame image. Supports durations of 5–15 seconds.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaad5f2/3kPTumERGySlFQrvJdc-5.jpg",
      "price": "Billing is calculated per second of video generated. Video at 768p costs $0.08 per second, with audio included. A 5-second video costs $0.40, and a 15-second video costs $1.20.",
      "license": "commercial",
      "published": "2026-09-21",
      "url": "https://fal.ai/models/minimax/h3-max/styles/hand-drawn"
    },
    {
      "id": "minimax/h3-max/styles/low-poly",
      "title": "H3 Max Low Poly",
      "category": "text-to-video",
      "family": "H3 Max",
      "description": "Generates 768p video with audio in a retro low-poly 3D style from text prompts or an optional first-frame image. Supports durations of 5–15 seconds.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaad5f8/7YH1svyhRf42Itvh-gQ4T.jpg",
      "price": "Billing is calculated per second of video generated. Video at 768p costs $0.08 per second, with audio included. A 5-second video costs $0.40, and a 15-second video costs $1.20.",
      "license": "commercial",
      "published": "2026-09-21",
      "url": "https://fal.ai/models/minimax/h3-max/styles/low-poly"
    },
    {
      "id": "minimax/h3-max/styles/retro-toon-70s",
      "title": "H3 Max Retro Toon 70s",
      "category": "text-to-video",
      "family": "H3 Max",
      "description": "Generates 768p video with audio in a retro 1970s hand-painted animation style from text prompts or an optional first-frame image. Supports durations of 5–15 seconds.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaad5fe/vIqC2SSPOPaR305av-iHV.jpg",
      "price": "Billing is calculated per second of video generated. Video at 768p costs $0.08 per second, with audio included. A 5-second video costs $0.40, and a 15-second video costs $1.20.",
      "license": "commercial",
      "published": "2026-09-21",
      "url": "https://fal.ai/models/minimax/h3-max/styles/retro-toon-70s"
    },
    {
      "id": "minimax/h3-max/styles/vhs",
      "title": "H3 Max VHS",
      "category": "text-to-video",
      "family": "H3 Max",
      "description": "Generates 768p VHS-style video with audio from text prompts or an optional first-frame image. Supports 5–15 second clips and adjustable tape damage, from subtle analog noise to strong tracking distortion.",
      "thumbnail": "https://v3b.fal.media/files/b/0aaad605/J4zSNIgmNAoArcg4BwPNp.jpg",
      "price": "Billing is calculated per second of video generated. Video at 768p costs $0.08 per second, with audio included. A 5-second video costs $0.40, and a 15-second video costs $1.20.",
      "license": "commercial",
      "published": "2026-09-21",
      "url": "https://fal.ai/models/minimax/h3-max/styles/vhs"
    },
    {
      "id": "tripo3d/p2/image-to-3d",
      "title": "Tripo P2 Image to 3D",
      "category": "image-to-3d",
      "family": "Tripo P2",
      "description": "Tripo P2 generates 3D models from a single image, with optional PBR textures, adjustable face counts, and triangle or quad mesh topology.",
      "thumbnail": "https://v3b.fal.media/files/b/0aab5c22/iR7ETXMnNL1Kgpj5bWU5k.jpg",
      "price": "Billing is calculated per generated 3D model. Each request costs $1.00 without textures, $1.10 with fast or standard textures, $1.20 with detailed textures, or $1.30 with extreme textures. PBR enables textures and uses the selected texture ",
      "license": "commercial",
      "published": "2026-09-21",
      "url": "https://fal.ai/models/tripo3d/p2/image-to-3d"
    },
    {
      "id": "tripo3d/p2/text-to-3d",
      "title": "Tripo P2 Text to 3D",
      "category": "text-to-3d",
      "family": "Tripo P2",
      "description": "Tripo P2 generates 3D models from a text prompt, with optional PBR textures, adjustable face counts, and triangle or quad mesh topology.",
      "thumbnail": "https://v3b.fal.media/files/b/0aab5c1f/wuw9kbrjl9AjbJDy54l0R.jpg",
      "price": "Billing is calculated per generated 3D model. Each request costs $1.00 without textures, $1.10 with fast or standard textures, $1.20 with detailed textures, or $1.30 with extreme textures. PBR enables textures and uses the selected texture ",
      "license": "commercial",
      "published": "2026-09-21",
      "url": "https://fal.ai/models/tripo3d/p2/text-to-3d"
    },
    {
      "id": "google/lyria-3.5",
      "title": "Lyria 3.5",
      "category": "text-to-audio",
      "family": "Lyria",
      "description": "Lyria 3.5 is Google DeepMind's latest music generation model, and you can generate almost any type of music with it",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa8afe/VbJWL93MNQf7eBuxjBEJO.jpg",
      "license": "commercial",
      "published": "2026-09-19",
      "url": "https://fal.ai/models/google/lyria-3.5"
    },
    {
      "id": "elevenlabs/music/v2",
      "title": "Elevenlabs Music v2",
      "category": "text-to-audio",
      "family": "ElevenLabs Music",
      "description": "Generate high quality, realistic music with fine controls using Elevenlabs Music v2!",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa6ac9/dGujDO5lw_IneNzkkfJBe.jpg",
      "price": "Your request will cost $0.6 per output audio minute. The audio will be rounded up to the closest minute. For instance, a generation with 30 seconds output will be billed as 1 minute.",
      "license": "commercial",
      "published": "2026-09-14",
      "url": "https://fal.ai/models/elevenlabs/music/v2"
    },
    {
      "id": "elevenlabs/music/v2.5",
      "title": "Elevenlabs Music v2.5",
      "category": "text-to-audio",
      "family": "ElevenLabs Music",
      "description": "Generate high quality, realistic music with fine controls using Elevenlabs Music v2.5!",
      "thumbnail": "https://v3b.fal.media/files/b/0aaa6adb/-Z3Ays503mRIC-atq70g8.jpg",
      "price": "Your request will cost $0.6 per output audio minute. The audio will be rounded up to the closest minute. For instance, a generation with 30 seconds output will be billed as 1 minute.",
      "license": "commercial",
      "published": "2026-09-14",
      "url": "https://fal.ai/models/elevenlabs/music/v2.5"
    },
    {
      "id": "fal-ai/elevenlabs/forced-alignment",
      "title": "Elevenlabs - Forced Alignment",
      "category": "speech-to-text",
      "description": "Align the transcript and your audio recording using Elevenlab's forced alignment feature!",
      "thumbnail": "https://v3b.fal.media/files/b/0aa999b9/uY0VJCvmmd7X0Y6-7CxK1.jpg",
      "price": "Your request will cost $0.22 per hour of audio input. It will be rounded upwards to the nearest hour.",
      "license": "commercial",
      "published": "2026-09-08",
      "url": "https://fal.ai/models/fal-ai/elevenlabs/forced-alignment"
    },
    {
      "id": "hitem3d/hi3d/v3.0/image-to-3d",
      "title": "Hi3D Image to 3D",
      "category": "image-to-3d",
      "family": "hi3dv3",
      "description": "Generate 3D models from a single image with Hi3D V3.0.",
      "thumbnail": "https://v3b.fal.media/files/b/0aa6fa71/Q-m26AkvFVMwHX7M3Mi4D.jpg",
      "price": "Your request is billed at $0.02 per credit. Total cost is a sum of geometry, texture, and PBR credits, where geometry credits scale with resolution, texture is fixed at 10 credits, and PBR is fixed at 5 credits. On v3.0, your request will c",
      "license": "commercial",
      "published": "2026-08-19",
      "url": "https://fal.ai/models/hitem3d/hi3d/v3.0/image-to-3d"
    },
    {
      "id": "hitem3d/hi3d/v3.0/multi-view-to-3d",
      "title": "Hi3D Multiview to 3D",
      "category": "image-to-3d",
      "family": "hi3dv3",
      "description": "Generate 3D models from multiple view images using Hi3D V3.0.",
      "thumbnail": "https://v3b.fal.media/files/b/0aa6fa9d/uNCL_9mZxOOwQUZ_W1zMI.jpg",
      "price": "Your request is billed at $0.02 per credit. Total cost is a sum of geometry, texture, and PBR credits, where geometry credits scale with resolution, texture is fixed at 10 credits, and PBR is fixed at 5 credits. On v3.0, your request will c",
      "license": "commercial",
      "published": "2026-08-19",
      "url": "https://fal.ai/models/hitem3d/hi3d/v3.0/multi-view-to-3d"
    },
    {
      "id": "hitem3d/hi3d/image-to-3d",
      "title": "Hi3D Image to 3D",
      "category": "image-to-3d",
      "family": "hi3d",
      "description": "Generate 3D models from a single image with Hi3D.",
      "thumbnail": "https://v3b.fal.media/files/b/0aa4770a/bTDPfMLZ2k4kj3D02kEkS.jpg",
      "price": "Your request is billed at $0.02 per credit. Total cost is a sum of geometry, texture, and PBR credits, where geometry credits scale with resolution, texture is fixed at 10 credits, and PBR (v2.0 and above) is fixed at 5 credits. On v1.5, yo",
      "license": "commercial",
      "published": "2026-08-06",
      "url": "https://fal.ai/models/hitem3d/hi3d/image-to-3d"
    },
    {
      "id": "hitem3d/hi3d/multi-view-to-3d",
      "title": "Hi3D Multiview to 3D",
      "category": "image-to-3d",
      "family": "hi3d",
      "description": "Generate 3D models from multiple view images using Hi3D.",
      "thumbnail": "https://v3b.fal.media/files/b/0aa47780/JV9CvgvE9dhJiWy68KhzF.jpg",
      "price": "Your request is billed at $0.02 per credit. Total cost is a sum of geometry, texture, and PBR credits, where geometry credits scale with resolution, texture is fixed at 10 credits, and PBR (v2.0 and above) is fixed at 5 credits. On v1.5, yo",
      "license": "commercial",
      "published": "2026-08-06",
      "url": "https://fal.ai/models/hitem3d/hi3d/multi-view-to-3d"
    },
    {
      "id": "alibaba/qwen-audio-3-tts",
      "title": "Qwen Audio 3.0 TTS (Flash)",
      "category": "text-to-speech",
      "family": "Qwen 3",
      "description": "Generate natural multilingual speech from text with fast voice and language control using Qwen Audio 3.0 TTS Flash.",
      "thumbnail": "https://v3b.fal.media/files/b/0aa349bc/YEHFJNX5VLA0_mQTdsH8F.jpg",
      "price": "",
      "license": "commercial",
      "published": "2026-07-29",
      "url": "https://fal.ai/models/alibaba/qwen-audio-3-tts"
    },
    {
      "id": "fal-ai/hyper3d/rodin/v2.5/fast",
      "title": "Hyper3D - Rodin V2.5 - Image to 3D - Fast",
      "category": "image-to-3d",
      "family": "Rodin 2.5",
      "description": "Rodin V2.5 by Hyper3D generates realistic and production ready 3D models from text or images. Do fast prototyping using the fast model.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9d154f/vAhd6ujr4i0bp9f238I9l_2901545ae364473781b3501c05e74f07.jpg",
      "price": "Your request will cost $0.1 per generation.",
      "license": "commercial",
      "published": "2026-06-19",
      "url": "https://fal.ai/models/fal-ai/hyper3d/rodin/v2.5/fast"
    },
    {
      "id": "fal-ai/hyper3d/rodin/v2.5/text-to-3d/fast",
      "title": "Hyper3D - Rodin V2.5 - Text to 3D - Fast",
      "category": "text-to-3d",
      "family": "Rodin 2.5",
      "description": "Rodin V2.5 by Hyper3D generates realistic and production ready 3D models from text or images. Do fast prototyping using the fast model.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9d156e/KgoS83QGNAiw4de9TFQGs_d1c95f9b763c452ea46048fbc905f6a2.jpg",
      "price": "Your request will cost $0.1 per generation.",
      "license": "commercial",
      "published": "2026-06-19",
      "url": "https://fal.ai/models/fal-ai/hyper3d/rodin/v2.5/text-to-3d/fast"
    },
    {
      "id": "async/tts-pro/v1.0",
      "title": "Async Text to Speech Pro V1.0",
      "category": "text-to-speech",
      "family": "Async TTS Pro",
      "description": "Generate professional-quality voiceovers in seconds with Async TTS Pro model text-based control over pauses, emphasis, and timing. Voice ids can be found at https://async.com/developer/voice-library",
      "thumbnail": "https://v3b.fal.media/files/b/0a9eca5b/cH54hF2DKMzegWqU-YP8X_c6ea3f36c9234aa9a137acf1b53cac01.jpg",
      "price": "You are going to be billed by the length of the input text, at a flat rate of $0.01 per 1,000 characters.",
      "license": "commercial",
      "published": "2026-06-18",
      "url": "https://fal.ai/models/async/tts-pro/v1.0"
    },
    {
      "id": "fal-ai/zonos2",
      "title": "Zonos2 Text to Speech",
      "category": "text-to-speech",
      "family": "Zonos",
      "description": "Zonos2 is a text-to-speech model that clones a voice from a short sample and speaks naturally across many languages.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9e7f9a/QwGx_S5uNbkzNRXsMZpDw_c64a2dc8166b4688a78bc80bfdb53ed0.jpg",
      "license": "commercial",
      "published": "2026-06-16",
      "url": "https://fal.ai/models/fal-ai/zonos2"
    },
    {
      "id": "nvidia/nemotron-asr-multilingual/asr",
      "title": "Nemotron Asr Multilingual",
      "category": "speech-to-text",
      "family": "Nemotron",
      "description": "Nemotron-ASR-Streaming is a multi lingual, streaming Automatic Speech Recognition (ASR) engineered to deliver high-quality multi lingual transcription across both low-latency streaming and high-throughput batch workloads.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9c2fbf/6-zNQDwpy54LnxDQx2dEO_a053aeaf6ad14a4fb4bce191086d95f1.jpg",
      "license": "commercial",
      "published": "2026-06-10",
      "url": "https://fal.ai/models/nvidia/nemotron-asr-multilingual/asr"
    },
    {
      "id": "fal-ai/bytedance/seed-speech/tts/v2",
      "title": "Bytedance Seed Speech Text to Speech",
      "category": "text-to-speech",
      "family": "Seed Speech",
      "description": "Seed Speech developed by ByteDance, is a family of large-scale text-to-speech models capable of synthesizing speech that is virtually indistinguishable from human speech.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9d0262/si1_Zv34axRlg3Uy9vzy__91af02a234d44b6485784f7e6c6c5f93.jpg",
      "license": "commercial",
      "published": "2026-06-05",
      "url": "https://fal.ai/models/fal-ai/bytedance/seed-speech/tts/v2"
    },
    {
      "id": "fal-ai/stable-audio-3/medium/base/audio-inpainting",
      "title": "Stable Audio 3 Medium Base Audio Inpainting",
      "category": "audio-to-audio",
      "family": "Stable Audio 3",
      "description": "Stable Audio 3 Medium Base audio inpainting is the foundational 1.4 billion parameter checkpoint for editing or filling selected stereo audio segments guided by text prompts.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9afeeb/bg-YuQIOPmZZgJj6OgW-k_f6720ee47c684abfb67105d0164385ba.jpg",
      "license": "commercial",
      "published": "2026-06-03",
      "url": "https://fal.ai/models/fal-ai/stable-audio-3/medium/base/audio-inpainting"
    },
    {
      "id": "fal-ai/stable-audio-3/medium/base/audio-outpainting",
      "title": "Stable Audio 3 Medium Base Audio Outpainting",
      "category": "audio-to-audio",
      "family": "Stable Audio 3",
      "description": "Stable Audio 3 Medium Base audio outpainting is the foundational 1.4 billion parameter checkpoint that extends existing stereo audio with causal continuation guided by text prompts.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9afee9/cOp5v1BcWtmgnpIxX8Sy7_450aa391c16e48bb9052769c7f293537.jpg",
      "license": "commercial",
      "published": "2026-06-03",
      "url": "https://fal.ai/models/fal-ai/stable-audio-3/medium/base/audio-outpainting"
    },
    {
      "id": "fal-ai/stable-audio-3/medium/base/audio-to-audio",
      "title": "Stable Audio 3 Medium Base Audio to Audio",
      "category": "audio-to-audio",
      "family": "Stable Audio 3",
      "description": "Stable Audio 3 Medium Base audio-to-audio is the foundational 1.4 billion parameter checkpoint that transforms input audio into new stereo variations up to 6 minutes guided by text prompts.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9afeee/Wj_uxy9sYVa6cMuxJueyT_0589558bdc874c7c838ed89bc6f4daaa.jpg",
      "license": "commercial",
      "published": "2026-06-03",
      "url": "https://fal.ai/models/fal-ai/stable-audio-3/medium/base/audio-to-audio"
    },
    {
      "id": "fal-ai/stable-audio-3/small/music/base/audio-outpainting",
      "title": "Stable Audio 3 Small Music Base Audio Outpainting",
      "category": "audio-to-audio",
      "family": "Stable Audio 3",
      "description": "Stable Audio 3 Small Music Base audio outpainting is the foundational 459 million parameter checkpoint that extends music tracks via causal continuation guided by text prompts.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9b9b0c/2gdxA9DEL1uyWlprJCdq8_fc0f082c9b354f6e9befbc9813d6b891.jpg",
      "license": "commercial",
      "published": "2026-06-03",
      "url": "https://fal.ai/models/fal-ai/stable-audio-3/small/music/base/audio-outpainting"
    },
    {
      "id": "fal-ai/stable-audio-3/small/sfx/base/audio-outpainting",
      "title": "Stable Audio 3 Small SFX Base Audio Outpainting",
      "category": "audio-to-audio",
      "family": "Stable Audio 3",
      "description": "Stable Audio 3 Small SFX Base audio outpainting is the foundational 459 million parameter checkpoint that extends sound-effect tracks via causal continuation guided by text prompts.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9b9afe/Qnmzr6trqXOOWM_QGjO7B_e9976fbcc60a415389b26ffe32ed091a.jpg",
      "license": "commercial",
      "published": "2026-06-03",
      "url": "https://fal.ai/models/fal-ai/stable-audio-3/small/sfx/base/audio-outpainting"
    },
    {
      "id": "fal-ai/hyper3d/rodin/v2.5/text-to-3d",
      "title": "Hyper3D - Rodin V2.5 - Text to 3D",
      "category": "text-to-3d",
      "family": "Rodin 2.5",
      "description": "Rodin V2.5 by Hyper3D generates realistic and production ready 3D models from text or images.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9c09f7/S7oXUUqZ5T_1m_3sQIQTZ_a9493c9cafed4ba1a43db14d36c9d243.jpg",
      "price": "Your request will cost $0.4 per generation. If HighPack is added to the `addons`, it will cost an additional $0.8 per generation.",
      "license": "commercial",
      "published": "2026-05-31",
      "url": "https://fal.ai/models/fal-ai/hyper3d/rodin/v2.5/text-to-3d"
    },
    {
      "id": "fal-ai/cohere-transcribe",
      "title": "Cohere Transcribe",
      "category": "speech-to-text",
      "family": "Cohere",
      "description": "Cohere Transcribe turns your business audio into accurate text, ready for search, analytics, and automation",
      "thumbnail": "https://v3b.fal.media/files/b/0a974bf5/gs1fpTrvtdkrKcBDh2gS8_ac4876872fb34e42a438031bb4fcdd00.jpg",
      "license": "commercial",
      "published": "2026-04-22",
      "url": "https://fal.ai/models/fal-ai/cohere-transcribe"
    },
    {
      "id": "fal-ai/gemini-3.1-flash-tts",
      "title": "Gemini 3.1 Flash Tts",
      "category": "text-to-speech",
      "family": "TTS",
      "description": "Newest audio model from Google introduces granular audio tags that give you precise control to direct AI speech for expressive audio generation.",
      "thumbnail": "https://v3b.fal.media/files/b/0a959ab9/uM2hZiKHOALvYZA2dBlB3_fa351c0ec002415e968d93c006094bae.jpg",
      "license": "commercial",
      "published": "2026-04-16",
      "url": "https://fal.ai/models/fal-ai/gemini-3.1-flash-tts"
    },
    {
      "id": "tripo3d/h3.1/text-to-3d",
      "title": "Tripo H3.1 Text to 3D",
      "category": "text-to-3d",
      "family": "Tripo H3.1",
      "description": "Generate 3D models from text descriptions using Tripo H3.1.",
      "thumbnail": "https://v3b.fal.media/files/b/0a95559a/M9VA3cYkl8K_pUsuQYiQz_992bf220fed24835892f1abe3027c871.jpg",
      "price": "Your request will cost $0.10 (without textures), $0.20 (with standard textures), or $0.30 (with HD textures), plus an additional $0.20 for detailed geometry and $0.05 for quad mesh if selected.",
      "license": "commercial",
      "published": "2026-04-15",
      "url": "https://fal.ai/models/tripo3d/h3.1/text-to-3d"
    },
    {
      "id": "tripo3d/p1/text-to-3d",
      "title": "Tripo P1 Text to 3D",
      "category": "text-to-3d",
      "family": "Tripo P1",
      "description": "Generate 3D models from text descriptions using Tripo P1.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9555d9/Iy0IDz5o0s5iyWHXMdnPQ_90e6f4ac24554af68acf654d79f36db3.jpg",
      "license": "commercial",
      "published": "2026-04-15",
      "url": "https://fal.ai/models/tripo3d/p1/text-to-3d"
    },
    {
      "id": "xai/tts/v1",
      "title": "xAI Text to Speech",
      "category": "text-to-speech",
      "family": "Grok TTS v1",
      "description": "Generate speech with expressive and realistic voices from xAI",
      "thumbnail": "https://v3b.fal.media/files/b/0a92763c/4toOv_gM9xUL_Zwmt4Gc8_67e9f077fc4c48e78dfa428f7d50b190.jpg",
      "price": "Your request will cost $0.015 per 1000 characters.",
      "license": "commercial",
      "published": "2026-03-17",
      "url": "https://fal.ai/models/xai/tts/v1"
    },
    {
      "id": "fal-ai/inworld-tts",
      "title": "Inworld TTS-1.5 Max",
      "category": "text-to-speech",
      "family": "TTS-1.5 Max",
      "description": "Text to Speech Endpoint for Inworld's TTS-1.5 Max.",
      "thumbnail": "https://v3b.fal.media/files/b/0a9207aa/h8_GEg6YVVRdB1ne848S5_0f9db25aa60c48e692aa3193466d0c7c.jpg",
      "price": "Your request will cost $0.01 per 1,000 characters.",
      "license": "commercial",
      "published": "2026-03-13",
      "url": "https://fal.ai/models/fal-ai/inworld-tts"
    },
    {
      "id": "fal-ai/meshy/v6/text-to-3d",
      "title": "Meshy 6",
      "category": "text-to-3d",
      "family": "Meshy 6",
      "description": "Meshy 6 generates 3D models from text prompts. It is an earlier standard model from Meshy; Meshy 7.1 is the latest standard model.",
      "thumbnail": "https://v3b.fal.media/files/b/0a8e1285/ItPdnJIwgzjDt76F_HWXf_6ac5e192c97e46cda9c929a829b786e4.jpg",
      "license": "commercial",
      "published": "2026-02-09",
      "url": "https://fal.ai/models/fal-ai/meshy/v6/text-to-3d"
    },
    {
      "id": "fal-ai/hunyuan-3d/v3.1/rapid/text-to-3d",
      "title": "Hunyuan 3d",
      "category": "text-to-3d",
      "family": "Hunyuan 3D",
      "description": "Create detailed, fully-textured 3D models with text",
      "thumbnail": "https://v3b.fal.media/files/b/0a8c542a/Qd0S35nuH5GWzvOH_eHjh_6812a7e9398647bab9255951a2b64306.jpg",
      "price": "Your request will cost $0.225 per generation. For $1.00, you can run this model 4 times. Base generation (text or image to 3D) costs $0.225. Enabling PBR materials adds $0.15.",
      "license": "commercial",
      "published": "2026-01-29",
      "url": "https://fal.ai/models/fal-ai/hunyuan-3d/v3.1/rapid/text-to-3d"
    },
    {
      "id": "fal-ai/hunyuan-3d/v3.1/pro/text-to-3d",
      "title": "Hunyuan 3D Pro Text to 3D",
      "category": "text-to-3d",
      "family": "Hunyuan 3D",
      "description": "Generate 3D models from text prompts with Hunyuan 3D Pro",
      "thumbnail": "https://v3b.fal.media/files/b/0a8c431d/UtISBa1ztdyamEsK0S2n7_49e72a0f05544a01b70bf20cebc2b795.jpg",
      "price": "Your request will cost $0.375 per generation. For $1.00, you can run this model 2 times. Enabling PBR materials adds $0.15. Using multi-view images adds $0.15. Custom face count adds $0.15.",
      "license": "commercial",
      "published": "2026-01-27",
      "url": "https://fal.ai/models/fal-ai/hunyuan-3d/v3.1/pro/text-to-3d"
    },
    {
      "id": "fal-ai/elevenlabs/speech-to-text/scribe-v2",
      "title": "ElevenLabs Speech to Text - Scribe V2",
      "category": "speech-to-text",
      "family": "Speech to Text",
      "description": "Use Scribe-V2 from ElevenLabs to do blazingly fast speech to text inferences!",
      "thumbnail": "https://v3b.fal.media/files/b/0a8a5d77/3XtHpcC46Sg0ItPRjB6Uk_82a3a831177745b38f925d35a7b5ed66.jpg",
      "price": "Your request will cost $0.008 per minute of input audio. If keyterm is used, your request will cost 30% more.",
      "license": "commercial",
      "published": "2026-01-14",
      "url": "https://fal.ai/models/fal-ai/elevenlabs/speech-to-text/scribe-v2"
    },
    {
      "id": "fal-ai/hunyuan-motion",
      "title": "Hunyuan Motion [1B]",
      "category": "text-to-3d",
      "family": "Hunyuan",
      "description": "Generate 3D human motions via text-to-generation interface of Hunyuan Motion!",
      "thumbnail": "https://v3b.fal.media/files/b/0a886121/DkVwzQN-tnXsEY9yW6wzM_473ed8506f41423181f1736357bde8f8.jpg",
      "license": "commercial",
      "published": "2025-12-30",
      "url": "https://fal.ai/models/fal-ai/hunyuan-motion"
    },
    {
      "id": "fal-ai/hunyuan-motion/fast",
      "title": "Hunyuan Motion [0.46B]",
      "category": "text-to-3d",
      "family": "Hunyuan",
      "description": "Generate 3D human motions via text-to-generation interface of Hunyuan Motion!",
      "thumbnail": "https://v3b.fal.media/files/b/0a886139/Y0oMLck1xMhSZq_SZO0kq_b33b912e9eb2401dbb38c48ec97b2cd3.jpg",
      "license": "commercial",
      "published": "2025-12-30",
      "url": "https://fal.ai/models/fal-ai/hunyuan-motion/fast"
    },
    {
      "id": "fal-ai/hunyuan3d-v3/text-to-3d",
      "title": "Hunyuan3d V3",
      "category": "text-to-3d",
      "family": "Hunyuan 3D 3",
      "description": "Turn simple sketches into detailed, fully-textured 3D models. Instantly convert your concept designs into formats ready for Unity, Unreal, and Blender.",
      "thumbnail": "https://v3b.fal.media/files/b/0a8696b5/YaGZtFEBFL6CEXjiVxdB__dbec569fa5d4418bbbc9d77354f7c86f.jpg",
      "price": "Billing is calculated per generation: $0.225 for Geometry, $0.375 for Normal, or $0.45 for LowPoly. Where available, PBR materials, multi-view images, and a custom face count each add $0.15. For example, a Normal generation with PBR materia",
      "license": "commercial",
      "published": "2025-12-16",
      "url": "https://fal.ai/models/fal-ai/hunyuan3d-v3/text-to-3d"
    },
    {
      "id": "fal-ai/speech-to-text",
      "title": "Speech-to-Text",
      "category": "speech-to-text",
      "description": "Leverage the rapid processing capabilities of AI models to enable accurate and efficient real-time speech-to-text transcription.",
      "thumbnail": "https://storage.googleapis.com/falserverless/gallery/canary.webp",
      "license": "commercial",
      "published": "2025-04-04",
      "url": "https://fal.ai/models/fal-ai/speech-to-text"
    },
    {
      "id": "fal-ai/speech-to-text/stream",
      "title": "Speech-To-text",
      "category": "speech-to-text",
      "description": "Leverage the rapid processing capabilities of AI models to enable accurate and efficient real-time speech-to-text transcription.",
      "thumbnail": "https://storage.googleapis.com/falserverless/gallery/canary.webp",
      "license": "commercial",
      "published": "2025-04-04",
      "url": "https://fal.ai/models/fal-ai/speech-to-text/stream"
    },
    {
      "id": "fal-ai/speech-to-text/turbo",
      "title": "Speech-to-Text",
      "category": "speech-to-text",
      "description": "Leverage the rapid processing capabilities of AI models to enable accurate and efficient real-time speech-to-text transcription.",
      "thumbnail": "https://storage.googleapis.com/falserverless/gallery/canary.webp",
      "license": "commercial",
      "published": "2025-04-04",
      "url": "https://fal.ai/models/fal-ai/speech-to-text/turbo"
    },
    {
      "id": "fal-ai/speech-to-text/turbo/stream",
      "title": "Speech-to-Text",
      "category": "speech-to-text",
      "description": "Leverage the rapid processing capabilities of AI models to enable accurate and efficient real-time speech-to-text transcription.",
      "thumbnail": "https://storage.googleapis.com/falserverless/gallery/canary.webp",
      "license": "commercial",
      "published": "2025-04-04",
      "url": "https://fal.ai/models/fal-ai/speech-to-text/turbo/stream"
    },
    {
      "id": "fal-ai/elevenlabs/speech-to-text",
      "title": "ElevenLabs Speech to Text",
      "category": "speech-to-text",
      "family": "Speech to Text",
      "description": "Generate text from speech using ElevenLabs advanced speech-to-text model.",
      "thumbnail": "https://storage.googleapis.com/falserverless/web-examples/elevenlabs/elevenlabs_thumbnail.webp",
      "license": "commercial",
      "published": "2025-02-27",
      "url": "https://fal.ai/models/fal-ai/elevenlabs/speech-to-text"
    },
    {
      "id": "fal-ai/wizper",
      "title": "Wizper (Whisper v3 -- fal.ai edition)",
      "category": "speech-to-text",
      "family": "Wizper",
      "description": "[Experimental] Whisper v3 Large -- but optimized by our inference wizards. Same WER, double the performance!",
      "thumbnail": "https://storage.googleapis.com/falserverless/gallery/wizper.webp",
      "license": "commercial",
      "published": "2024-04-08",
      "url": "https://fal.ai/models/fal-ai/wizper"
    }
  ],
  "orMedia": [
    {
      "id": "00/alibaba/happyhorse-1.0",
      "orId": "alibaba/happyhorse-1.0",
      "kind": "video",
      "label": "Alibaba: HappyHorse 1.0",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "HappyHorse 1.0 is a video generation model from Alibaba. It generates short videos from a text prompt, a single starting image, or a set of reference images, with output up...",
      "released": "2026-06-24",
      "durations": [
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "720p",
        "1080p"
      ],
      "price": "$0.198 per second of video at 720p, up to $0.339/s at 1080p, on your AI tokens.",
      "priceShort": "$0.198 / s (720p)"
    },
    {
      "id": "00/alibaba/happyhorse-1.1",
      "orId": "alibaba/happyhorse-1.1",
      "kind": "video",
      "label": "Alibaba: HappyHorse 1.1",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "HappyHorse 1.1 is a video generation model from Alibaba. It generates short videos from a text prompt, a single starting image, or a set of reference images, with output up...",
      "released": "2026-06-24",
      "durations": [
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "720p",
        "1080p"
      ],
      "price": "$0.198 per second of video at 720p, up to $0.256/s at 1080p, on your AI tokens.",
      "priceShort": "$0.198 / s (720p)"
    },
    {
      "id": "00/alibaba/wan-2.6",
      "orId": "alibaba/wan-2.6",
      "kind": "video",
      "label": "Alibaba: Wan 2.6",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Alibaba's most advanced video generation model, supporting over 10 visual creation capabilities in a unified system. Wan 2.6 generates 1080p video at 24fps from text, images, reference videos, or audio,...",
      "released": "2026-03-28",
      "durations": [
        5,
        10
      ],
      "resolutions": [
        "720p",
        "1080p"
      ],
      "price": "$0.08 per second of video (text to video, 480p), up to $0.3/s (image to video, 1080p), on your AI tokens.",
      "priceShort": "$0.08 / s (text to video, 480p)"
    },
    {
      "id": "00/alibaba/wan-2.7",
      "orId": "alibaba/wan-2.7",
      "kind": "video",
      "label": "Alibaba: Wan 2.7",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Wan 2.7 is a video generation model from Alibaba. It supports text-to-video, image-to-video with first and last frame control, and reference-to-video, where multiple reference images guide the style and content...",
      "released": "2026-04-15",
      "durations": [
        2,
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10
      ],
      "resolutions": [
        "720p",
        "1080p"
      ],
      "price": "$0.2 per second of video, on your AI tokens.",
      "priceShort": "$0.2 / s"
    },
    {
      "id": "00/alibaba/wan-3.0",
      "orId": "alibaba/wan-3.0",
      "kind": "video",
      "label": "Alibaba: Wan 3.0",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Wan 3.0 is a video generation model from Alibaba for text-to-video, image-to-video, and reference-guided video generation. It produces 480p, 720p, or 1080p video with durations from 2 to 30 seconds.",
      "released": "2026-08-24",
      "durations": [
        2,
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15,
        16,
        17,
        18,
        19,
        20,
        21,
        22,
        23,
        24,
        25,
        26,
        27,
        28,
        29,
        30
      ],
      "resolutions": [
        "480p",
        "720p",
        "1080p"
      ],
      "price": "$0.1 per second of video at 480p, up to $0.4/s at 1080p, on your AI tokens.",
      "priceShort": "$0.1 / s (480p)"
    },
    {
      "id": "00/alibaba/wan-3.0-prime",
      "orId": "alibaba/wan-3.0-prime",
      "kind": "video",
      "label": "Alibaba: Wan 3.0 Prime",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Wan 3.0 Prime is a fast-mode variant of Wan 3.0 from Alibaba. It supports text-to-video and first-frame image-to-video generation.",
      "released": "2026-08-27",
      "durations": [
        2,
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15,
        16,
        17,
        18,
        19,
        20,
        21,
        22,
        23,
        24,
        25,
        26,
        27,
        28,
        29,
        30
      ],
      "resolutions": [
        "480p",
        "720p",
        "1080p"
      ],
      "price": "$0.136 per second of video at 480p, up to $0.56/s at 1080p, on your AI tokens.",
      "priceShort": "$0.136 / s (480p)"
    },
    {
      "id": "00/black-forest-labs/flux-3-image",
      "orId": "black-forest-labs/flux-3-image",
      "kind": "image",
      "label": "Black Forest Labs: FLUX 3 Image",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "FLUX.3 Image is Black Forest Labs' flagship image generation and editing model. It handles text-to-image and multi-reference editing with up to 10 input images, and renders at fixed resolution tiers...",
      "released": "2026-10-01",
      "price": "$0.082 per image, on your AI tokens.",
      "priceShort": "$0.082 / image"
    },
    {
      "id": "00/black-forest-labs/flux-3-image-klein",
      "orId": "black-forest-labs/flux-3-image-klein",
      "kind": "image",
      "label": "Black Forest Labs: FLUX 3 Image Klein",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "FLUX 3 Image Klein is an open-weights image generation and editing model from Black Forest Labs. It is suited for text-to-image, multi-reference editing that keeps characters and subjects consistent across...",
      "released": "2026-10-15",
      "price": "$0.016 per image, on your AI tokens.",
      "priceShort": "$0.016 / image"
    },
    {
      "id": "00/black-forest-labs/flux-3-video",
      "orId": "black-forest-labs/flux-3-video",
      "kind": "video",
      "label": "Black Forest Labs: FLUX 3 Video",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "FLUX.3 Video is a video generation model from Black Forest Labs. It supports text-to-video, image-guided generation with opening and closing keyframes, and video continuation workflows, making it suited for controlled...",
      "released": "2026-08-04",
      "durations": [
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15,
        16,
        17,
        18,
        19,
        20
      ],
      "resolutions": [
        "720p",
        "1080p",
        "2K",
        "4K"
      ],
      "price": "$0.34 per second of video, up to $0.82/s at 4k, on your AI tokens.",
      "priceShort": "$0.34 / s"
    },
    {
      "id": "00/black-forest-labs/flux-video-edit",
      "orId": "black-forest-labs/flux-video-edit",
      "kind": "video",
      "label": "Black Forest Labs: FLUX Video Edit",
      "modes": [
        "video→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "FLUX Video Edit [fast] takes a source video and an edit prompt and returns a precisely edited video. Add, remove, or replace objects and characters, rebuild the setting, edit on-screen...",
      "released": "2026-09-10",
      "price": "$0.06 per second of video, on your AI tokens.",
      "priceShort": "$0.06 / s"
    },
    {
      "id": "00/black-forest-labs/flux-video-upscale",
      "orId": "black-forest-labs/flux-video-upscale",
      "kind": "video",
      "label": "Black Forest Labs: FLUX Video Upscale",
      "modes": [
        "video→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "FLUX Video Upscale is a video upscaling model from Black Forest Labs. It enlarges a single source video by 1.5× to 3× while preserving its duration, with an optional prompt...",
      "released": "2026-08-19",
      "price": "$0.07 per megapixel-second (precise), up to $0.1/MP·s (creative), on your AI tokens.",
      "priceShort": "$0.07 / MP·s (precise)"
    },
    {
      "id": "00/black-forest-labs/flux.2-flex",
      "orId": "black-forest-labs/flux.2-flex",
      "kind": "image",
      "label": "Black Forest Labs: FLUX.2 Flex",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "FLUX.2 [flex] excels at rendering complex text, typography, and fine details, and supports multi-reference editing in the same unified architecture. Pricing is as follows, [per the docs](https://bfl.ai/pricing?category=flux.2): We charge $0.06...",
      "released": "2025-11-25",
      "price": "$0.12 per megapixel, on your AI tokens.",
      "priceShort": "$0.12 / megapixel"
    },
    {
      "id": "00/black-forest-labs/flux.2-klein-4b",
      "orId": "black-forest-labs/flux.2-klein-4b",
      "kind": "image",
      "label": "Black Forest Labs: FLUX.2 Klein 4B",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "FLUX.2 [klein] 4B is the fastest and most cost-effective model in the FLUX.2 family, optimized for high-throughput use cases while maintaining excellent image quality. Pricing is based on the output...",
      "released": "2026-01-14",
      "price": "$0.028 per megapixel, on your AI tokens.",
      "priceShort": "$0.028 / megapixel"
    },
    {
      "id": "00/black-forest-labs/flux.2-max",
      "orId": "black-forest-labs/flux.2-max",
      "kind": "image",
      "label": "Black Forest Labs: FLUX.2 Max",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "FLUX.2 [max] is the new top-tier image model from Black Forest Labs, pushing image quality, prompt understanding, and editing consistency to the highest level yet. Pricing is as follows, [per...",
      "released": "2025-12-16",
      "price": "$0.14 per megapixel, on your AI tokens.",
      "priceShort": "$0.14 / megapixel"
    },
    {
      "id": "00/black-forest-labs/flux.2-pro",
      "orId": "black-forest-labs/flux.2-pro",
      "kind": "image",
      "label": "Black Forest Labs: FLUX.2 Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "A high-end image generation and editing model focused on frontier-level visual quality and reliability. It delivers strong prompt adherence, stable lighting, sharp textures, and consistent character/style reproduction across multi-reference inputs....",
      "released": "2025-11-25",
      "price": "$0.06 per megapixel, on your AI tokens.",
      "priceShort": "$0.06 / megapixel"
    },
    {
      "id": "00/bytedance-seed/seedream-4.5",
      "orId": "bytedance-seed/seedream-4.5",
      "kind": "image",
      "label": "ByteDance Seed: Seedream 4.5",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Seedream 4.5 is the latest in-house image generation model developed by ByteDance. Compared with Seedream 4.0, it delivers comprehensive improvements, especially in editing consistency, including better preservation of subject details,...",
      "released": "2025-12-23",
      "price": "$0.08 per image, on your AI tokens.",
      "priceShort": "$0.08 / image"
    },
    {
      "id": "00/bytedance-seed/seedream-5-0-flash",
      "orId": "bytedance-seed/seedream-5-0-flash",
      "kind": "image",
      "label": "ByteDance Seed: Seedream 5.0 Flash",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Seedream 5.0 Flash is an image generation and editing model from ByteDance Seed. It is the fast, cost-efficient tier of the Seedream 5.0 family, suited for high-volume production and interactive...",
      "released": "2026-10-01",
      "price": "$0.036 per image, on your AI tokens.",
      "priceShort": "$0.036 / image"
    },
    {
      "id": "00/bytedance-seed/seedream-5-0-lite",
      "orId": "bytedance-seed/seedream-5-0-lite",
      "kind": "image",
      "label": "ByteDance Seed: Seedream 5.0 Lite",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Seedream 5.0 Lite is an image generation model from ByteDance Seed. It is suited for professional visual creation that benefits from web-connected retrieval, complex-prompt comprehension, visual references, and broad knowledge...",
      "released": "2026-08-13",
      "price": "$0.07 per image, on your AI tokens.",
      "priceShort": "$0.07 / image"
    },
    {
      "id": "00/bytedance-seed/seedream-5-0-pro",
      "orId": "bytedance-seed/seedream-5-0-pro",
      "kind": "image",
      "label": "ByteDance Seed: Seedream 5.0 Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Seedream 5.0 Pro is an image generation and editing model from ByteDance Seed. It is suited for commercial visual-production workflows that require precise editing control, lifelike scenes, and natural rendering.",
      "released": "2026-08-12",
      "price": "$0.09 per image, on your AI tokens.",
      "priceShort": "$0.09 / image"
    },
    {
      "id": "00/bytedance-seed/seedream-5-5",
      "orId": "bytedance-seed/seedream-5-5",
      "kind": "image",
      "label": "ByteDance Seed: Seedream 5.5",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Seedream 5.5 is an image generation and editing model from ByteDance Seed. It is suited for creating images from text prompts and for editing or combining existing images, using up...",
      "released": "2026-11-12",
      "price": "$0.04 per image, on your AI tokens.",
      "priceShort": "$0.04 / image"
    },
    {
      "id": "00/bytedance/seedance-2.0",
      "orId": "bytedance/seedance-2.0",
      "kind": "video",
      "label": "ByteDance: Seedance 2.0",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Seedance 2.0 is a video generation model from ByteDance. It supports text-to-video, image-to-video with first and last frame control, and multimodal reference-to-video. It is particularly strong at preserving character consistency,...",
      "released": "2026-04-15",
      "durations": [
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "480p",
        "720p",
        "1080p",
        "4K"
      ],
      "price": "$4.80 per million video tokens (4k with video input), up to $15.40/M video tokens at 1080p, on your AI tokens.",
      "priceShort": "$4.80 / M video tokens (4k with video input)"
    },
    {
      "id": "00/bytedance/seedance-2.0-fast",
      "orId": "bytedance/seedance-2.0-fast",
      "kind": "video",
      "label": "ByteDance: Seedance 2.0 Fast",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Seedance 2.0 Fast is a video generation model from ByteDance. It supports text-to-video, image-to-video with first and last frame control, and multimodal reference-to-video. It prioritizes generation speed and lower cost...",
      "released": "2026-04-15",
      "durations": [
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "480p",
        "720p"
      ],
      "price": "$4.95 per million video tokens (with video input), up to $8.40/M video tokens (without audio), on your AI tokens.",
      "priceShort": "$4.95 / M video tokens (with video input)"
    },
    {
      "id": "00/bytedance/seedance-2.0-mini",
      "orId": "bytedance/seedance-2.0-mini",
      "kind": "video",
      "label": "ByteDance: Seedance 2.0 Mini",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Seedance 2.0 Mini is a video generation model from ByteDance. It supports text-to-video, image-to-video with first and last frame control, and multimodal reference-to-video with image, video, and audio inputs. It...",
      "released": "2026-08-12",
      "durations": [
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "480p",
        "720p"
      ],
      "price": "$4.20 per million video tokens (with video input), up to $7/M video tokens (without audio), on your AI tokens.",
      "priceShort": "$4.20 / M video tokens (with video input)"
    },
    {
      "id": "00/bytedance/seedance-2.5",
      "orId": "bytedance/seedance-2.5",
      "kind": "video",
      "label": "ByteDance: Seedance 2.5",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Seedance 2.5 is a video generation model from ByteDance. It is suited for long-form storytelling, multimodal reference-based generation, video editing, and video extension. It supports first-frame and first-and-last-frame control, up...",
      "released": "2026-08-07",
      "durations": [
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15,
        16,
        17,
        18,
        19,
        20,
        21,
        22,
        23,
        24,
        25,
        26,
        27,
        28,
        29,
        30
      ],
      "resolutions": [
        "480p",
        "720p"
      ],
      "price": "$12.80 per million video tokens (with video input), up to $21.40/M video tokens (without audio), on your AI tokens.",
      "priceShort": "$12.80 / M video tokens (with video input)"
    },
    {
      "id": "00/google/gemini-2.5-flash-image",
      "orId": "google/gemini-2.5-flash-image",
      "kind": "image",
      "label": "Google: Nano Banana (Gemini 2.5 Flash Image)",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Gemini 2.5 Flash Image, a.k.a. \"Nano Banana,\" is now generally available. It is a state of the art image generation model with contextual understanding. It is capable of image generation,...",
      "released": "2025-10-07",
      "price": "≈ $0.077 per image, on your AI tokens.",
      "priceShort": "≈ $0.077 / image"
    },
    {
      "id": "00/google/gemini-3-pro-image",
      "orId": "google/gemini-3-pro-image",
      "kind": "image",
      "label": "Google: Nano Banana Pro (Gemini 3 Pro Image)",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Nano Banana Pro is Google’s most advanced image-generation and editing model, built on Gemini 3 Pro. It extends the original Nano Banana with significantly improved multimodal reasoning, real-world grounding, and...",
      "released": "2026-06-18",
      "price": "≈ $0.31 per image, on your AI tokens.",
      "priceShort": "≈ $0.31 / image"
    },
    {
      "id": "00/google/gemini-3-pro-image-preview",
      "orId": "google/gemini-3-pro-image-preview",
      "kind": "image",
      "label": "Google: Nano Banana Pro (Gemini 3 Pro Image Preview)",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Nano Banana Pro is Google’s most advanced image-generation and editing model, built on Gemini 3 Pro. It extends the original Nano Banana with significantly improved multimodal reasoning, real-world grounding, and...",
      "released": "2025-11-20",
      "price": "≈ $0.155 per image, on your AI tokens.",
      "priceShort": "≈ $0.155 / image"
    },
    {
      "id": "00/google/gemini-3.1-flash-image",
      "orId": "google/gemini-3.1-flash-image",
      "kind": "image",
      "label": "Google: Nano Banana 2 (Gemini 3.1 Flash Image)",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Gemini 3.1 Flash Image, a.k.a. \"Nano Banana 2,\" is Google’s latest state of the art image generation and editing model, delivering Pro-level visual quality at Flash speed. It combines advanced...",
      "released": "2026-06-18",
      "price": "≈ $0.17 per image, on your AI tokens.",
      "priceShort": "≈ $0.17 / image"
    },
    {
      "id": "00/google/gemini-3.1-flash-image-preview",
      "orId": "google/gemini-3.1-flash-image-preview",
      "kind": "image",
      "label": "Google: Nano Banana 2 (Gemini 3.1 Flash Image Preview)",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Gemini 3.1 Flash Image Preview, a.k.a. \"Nano Banana 2,\" is Google’s latest state of the art image generation and editing model, delivering Pro-level visual quality at Flash speed. It combines...",
      "released": "2026-02-26",
      "price": "≈ $0.155 per image, on your AI tokens.",
      "priceShort": "≈ $0.155 / image"
    },
    {
      "id": "00/google/gemini-3.1-flash-lite-image",
      "orId": "google/gemini-3.1-flash-lite-image",
      "kind": "image",
      "label": "Google: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) is Google's fastest, most cost-efficient Gemini image model, built for high-velocity developer pipelines and rapid-fire visual exploration. It delivers text-to-image generation...",
      "released": "2026-06-30",
      "price": "≈ $0.077 per image, on your AI tokens.",
      "priceShort": "≈ $0.077 / image"
    },
    {
      "id": "00/google/gemini-nano-banana-2.1",
      "orId": "google/gemini-nano-banana-2.1",
      "kind": "image",
      "label": "Google: Nano Banana 2.1",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Nano Banana 2.1 (Gemini Nano Banana 2.1) is Google's image generation and editing model on the Flash tier, succeeding Nano Banana 2 and Nano Banana Pro. It improves product recontextualization,...",
      "released": "2026-10-06",
      "price": "≈ $0.077 per image, on your AI tokens.",
      "priceShort": "≈ $0.077 / image"
    },
    {
      "id": "00/google/gemini-omni-1.1-flash",
      "orId": "google/gemini-omni-1.1-flash",
      "kind": "video",
      "label": "Google: Gemini Omni 1.1 Flash",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Gemini Omni 1.1 Flash is Google's native multimodal video generation and editing model, the generally available successor to Gemini Omni Flash Preview. It is suited for text-to-video, image-to-video, first and...",
      "released": "2026-11-04",
      "durations": [
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10
      ],
      "resolutions": [
        "720p",
        "1080p",
        "4K",
        "360p"
      ],
      "price": "$0.068 per second of video (with audio 360p), up to $0.608/s (with audio 4k), on your AI tokens.",
      "priceShort": "$0.068 / s (with audio 360p)"
    },
    {
      "id": "00/google/gemini-omni-1.2-flash",
      "orId": "google/gemini-omni-1.2-flash",
      "kind": "video",
      "label": "Google: Gemini Omni 1.2 Flash",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Gemini Omni 1.2 Flash is Google's native multimodal video generation and editing model, the successor to Gemini Omni 1.1 Flash. It generates clips up to 15 seconds long with native...",
      "released": "2026-11-05",
      "durations": [
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "720p",
        "1080p",
        "4K",
        "360p"
      ],
      "price": "$0.046 per second of video (with audio 360p), up to $0.417/s (with audio 4k), on your AI tokens.",
      "priceShort": "$0.046 / s (with audio 360p)"
    },
    {
      "id": "00/google/veo-3.1",
      "orId": "google/veo-3.1",
      "kind": "video",
      "label": "Google: Veo 3.1",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Google's state-of-the-art video generation model, built for maximum visual fidelity in final production cuts. Veo 3.1 generates high-quality 1080p video from text or image prompts with native synchronized audio —...",
      "released": "2026-03-23",
      "durations": [
        4,
        6,
        8
      ],
      "resolutions": [
        "720p",
        "1080p",
        "4K"
      ],
      "price": "$0.4 per second of video (without audio), up to $1.20/s (with audio 4k), on your AI tokens.",
      "priceShort": "$0.4 / s (without audio)"
    },
    {
      "id": "00/google/veo-3.1-fast",
      "orId": "google/veo-3.1-fast",
      "kind": "video",
      "label": "Google: Veo 3.1 Fast",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Google's mid-tier video generation model balancing speed and quality. Veo 3.1 Fast generates high-quality video from text or image prompts with native synchronized audio, offering faster turnaround than Veo 3.1...",
      "released": "2026-04-24",
      "durations": [
        4,
        6,
        8
      ],
      "resolutions": [
        "720p",
        "1080p",
        "4K"
      ],
      "price": "$0.16 per second of video (without audio 720p), up to $0.6/s (with audio 4k), on your AI tokens.",
      "priceShort": "$0.16 / s (without audio 720p)"
    },
    {
      "id": "00/google/veo-3.1-lite",
      "orId": "google/veo-3.1-lite",
      "kind": "video",
      "label": "Google: Veo 3.1 Lite",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Google's most cost-effective video generation model, designed for high-volume applications and rapid iteration. Veo 3.1 Lite generates 720p and 1080p video from text or image prompts with native synchronized audio...",
      "released": "2026-04-23",
      "durations": [
        4,
        6,
        8
      ],
      "resolutions": [
        "720p",
        "1080p"
      ],
      "price": "$0.06 per second of video (without audio 720p), up to $0.16/s (with audio), on your AI tokens.",
      "priceShort": "$0.06 / s (without audio 720p)"
    },
    {
      "id": "00/heygen/avatar-iv",
      "orId": "heygen/avatar-iv",
      "kind": "video",
      "label": "HeyGen: Avatar IV",
      "modes": [
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "HeyGen: Avatar IV is an image-to-video model that animates a single photo into an expressive, lip-synced talking-head video. Rather than only matching mouth shapes to words, it interprets the vocal...",
      "released": "2026-08-24",
      "resolutions": [
        "720p",
        "1080p"
      ],
      "price": "$0.1 per second of video, on your AI tokens.",
      "priceShort": "$0.1 / s"
    },
    {
      "id": "00/heygen/heygen-video-1",
      "orId": "heygen/heygen-video-1",
      "kind": "video",
      "label": "HeyGen: HeyGen Video",
      "modes": [
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "HeyGen Video is a general-purpose video generation model from HeyGen. It renders short clips with synthesized audio (dialogue, ambience, and sound effects) in a single call, working from a text...",
      "released": "2026-09-30",
      "durations": [
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15,
        16,
        17,
        18,
        19,
        20,
        21,
        22,
        23,
        24,
        25,
        26,
        27,
        28,
        29,
        30
      ],
      "resolutions": [
        "480p",
        "768p",
        "1080p",
        "2K"
      ],
      "price": "$0.04 per second of video at 480p, up to $0.2/s at 1080p, on your AI tokens.",
      "priceShort": "$0.04 / s (480p)"
    },
    {
      "id": "00/krea/k3-edit",
      "orId": "krea/k3-edit",
      "kind": "image",
      "label": "Krea: K3 Edit",
      "modes": [
        "image→image"
      ],
      "kinds": [
        "image-edit"
      ],
      "description": "Krea K3 Edit is an image-to-image editing model from Krea. It takes one to eight reference images plus a text prompt and returns a single edited image, following instructions that...",
      "released": "2026-10-15"
    },
    {
      "id": "00/krea/krea-2-large",
      "orId": "krea/krea-2-large",
      "kind": "image",
      "label": "Krea: Krea 2 Large",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Krea 2 Large is Krea's high-capability image generation model, more than twice the size of Krea 2 Medium. Its lighter post-training gives images a rawer, more textured, and flexible character,...",
      "released": "2026-07-20",
      "price": "$0.12 per image, on your AI tokens.",
      "priceShort": "$0.12 / image"
    },
    {
      "id": "00/krea/krea-2-medium",
      "orId": "krea/krea-2-medium",
      "kind": "image",
      "label": "Krea: Krea 2 Medium",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Krea 2 Medium is Krea's balanced, cost-efficient image generation model and a practical starting point for a broad range of use cases. Its extensive post-training supports stable, consistent generations, with...",
      "released": "2026-07-20",
      "price": "$0.06 per image, on your AI tokens.",
      "priceShort": "$0.06 / image"
    },
    {
      "id": "00/krea/krea-2-medium-turbo",
      "orId": "krea/krea-2-medium-turbo",
      "kind": "image",
      "label": "Krea: Krea 2 Medium Turbo",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Krea 2 Medium Turbo is a distilled, speed-focused variant of Krea 2 Medium from Krea. It is designed for rapid iteration and graphic design exploration where fast generation is the...",
      "released": "2026-07-20",
      "price": "$0.03 per image, on your AI tokens.",
      "priceShort": "$0.03 / image"
    },
    {
      "id": "00/kwaivgi/kling-v3.0-pro",
      "orId": "kwaivgi/kling-v3.0-pro",
      "kind": "video",
      "label": "Kling: Video v3.0 Pro",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Kling v3.0 Pro is Kuaishou's premium video generation model, offering higher visual quality than the Standard tier. It supports text-to-video and image-to-video workflows, with first-frame and last-frame control for precise...",
      "released": "2026-04-29",
      "durations": [
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "720p"
      ],
      "price": "$0.224 per second of video, up to $0.336/s (with audio), on your AI tokens.",
      "priceShort": "$0.224 / s"
    },
    {
      "id": "00/kwaivgi/kling-v3.0-std",
      "orId": "kwaivgi/kling-v3.0-std",
      "kind": "video",
      "label": "Kling: Video v3.0 Standard",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Kling v3.0 Standard is a video generation model from Kuaishou. It supports text-to-video and image-to-video workflows, with first-frame and last-frame control for guided scene composition. Clips range from 3 to...",
      "released": "2026-04-29",
      "durations": [
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "720p"
      ],
      "price": "$0.168 per second of video, up to $0.252/s (with audio), on your AI tokens.",
      "priceShort": "$0.168 / s"
    },
    {
      "id": "00/kwaivgi/kling-video-o1",
      "orId": "kwaivgi/kling-video-o1",
      "kind": "video",
      "label": "Kling: Video O1",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Kling Video O1 is a video generation model from Kuaishou. It supports text and image inputs with video output, enabling text-to-video and image-to-video workflows. It is suited for cinematic content...",
      "released": "2026-04-20",
      "durations": [
        5,
        10
      ],
      "resolutions": [
        "720p"
      ],
      "price": "$0.224 per second of video, on your AI tokens.",
      "priceShort": "$0.224 / s"
    },
    {
      "id": "00/microsoft/mai-image-2.5",
      "orId": "microsoft/mai-image-2.5",
      "kind": "image",
      "label": "Microsoft AI: MAI-Image-2.5",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Microsoft AI's MAI-Image-2.5 is a high-quality image generation model available via Azure AI Foundry. It produces photorealistic and artistic images from text prompts with support for various aspect ratios.",
      "released": "2026-06-02",
      "price": "≈ $0.121 per image, on your AI tokens.",
      "priceShort": "≈ $0.121 / image"
    },
    {
      "id": "00/microsoft/mai-image-2.5-pro",
      "orId": "microsoft/mai-image-2.5-pro",
      "kind": "image",
      "label": "Microsoft AI: MAI-Image-2.5 Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Microsoft AI's MAI-Image-2.5 is a high-quality image generation model available via Azure AI Foundry. It produces photorealistic and artistic images from text prompts with support for various aspect ratios.",
      "released": "2026-07-23",
      "price": "≈ $0.279 per image, on your AI tokens.",
      "priceShort": "≈ $0.279 / image"
    },
    {
      "id": "00/microsoft/mai-image-2.6",
      "orId": "microsoft/mai-image-2.6",
      "kind": "image",
      "label": "Microsoft AI: MAI-Image-2.6",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "MAI-Image-2.6 is an image generation and editing model from Microsoft AI, the precision tier of the MAI-Image-2.6 family alongside the faster [MAI-Image-2.6 Flash](/microsoft/mai-image-2.6-flash). It is suited for design-ready visuals and...",
      "released": "2026-09-04",
      "price": "≈ $0.098 per image, on your AI tokens.",
      "priceShort": "≈ $0.098 / image"
    },
    {
      "id": "00/microsoft/mai-image-2.6-flash",
      "orId": "microsoft/mai-image-2.6-flash",
      "kind": "image",
      "label": "Microsoft AI: MAI-Image-2.6 Flash",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "MAI-Image-2.6 Flash is the lower-latency, lower-cost member of the [MAI-Image-2.6](/microsoft/mai-image-2.6) family from Microsoft AI, built for latency-sensitive, high-throughput production image generation and editing at comparable quality to the precision tier....",
      "released": "2026-09-04",
      "price": "≈ $0.049 per image, on your AI tokens.",
      "priceShort": "≈ $0.049 / image"
    },
    {
      "id": "00/minimax/hailuo-2.3",
      "orId": "minimax/hailuo-2.3",
      "kind": "video",
      "label": "MiniMax: Hailuo 2.3",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Hailuo 2.3 is a video generation model from MiniMax. It accepts text prompts and reference images as input and generates video output, supporting both text-to-video and image-to-video workflows. It is...",
      "released": "2026-04-20",
      "durations": [
        6,
        10
      ],
      "resolutions": [
        "1080p"
      ],
      "price": "$0.163 per second of video, on your AI tokens.",
      "priceShort": "$0.163 / s"
    },
    {
      "id": "00/minimax/hailuo-3",
      "orId": "minimax/hailuo-3",
      "kind": "video",
      "label": "MiniMax: H3",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "MiniMax H3 is a lightweight, open-weights video generation model from MiniMax. It is designed for precise multimodal editing and controlled content generation, including instruction-guided edits, text and brand rendering, and...",
      "released": "2026-07-29",
      "durations": [
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "2K"
      ],
      "price": "$0.26 per second of video, on your AI tokens.",
      "priceShort": "$0.26 / s"
    },
    {
      "id": "00/minimax/hailuo-3-max",
      "orId": "minimax/hailuo-3-max",
      "kind": "video",
      "label": "MiniMax: H3 Max",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "MiniMax H3 Max is a video-generation model from MiniMax, jointly released with fal.ai. Derived through additional training from MiniMax H3, it is designed for faster text-to-video and image-to-video generation with...",
      "released": "2026-09-02",
      "durations": [
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "768p",
        "480p"
      ],
      "price": "$0.1 per second of video at 480p, up to $0.16/s at 768p, on your AI tokens.",
      "priceShort": "$0.1 / s (480p)"
    },
    {
      "id": "00/minimax/minimax-h3-image",
      "orId": "minimax/minimax-h3-image",
      "kind": "image",
      "label": "MiniMax: MiniMax H3 Image",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "MiniMax H3 Image is MiniMax's image generation and editing model. It creates images from text prompts and edits or composes images from 1 to 16 reference images, with output up...",
      "released": "2026-11-10",
      "price": "≈ $0.01 per image, on your AI tokens.",
      "priceShort": "≈ $0.01 / image"
    },
    {
      "id": "00/minimax/minimax-h3-image-highspeed",
      "orId": "minimax/minimax-h3-image-highspeed",
      "kind": "image",
      "label": "MiniMax: MiniMax H3 Image Highspeed",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "MiniMax H3 Image Highspeed is the faster version of MiniMax's H3 Image model. It takes the same inputs and limits as MiniMax H3 Image: it creates images from text prompts...",
      "released": "2026-11-11",
      "price": "≈ $0.02 per image, on your AI tokens.",
      "priceShort": "≈ $0.02 / image"
    },
    {
      "id": "00/openai/gpt-5-image",
      "orId": "openai/gpt-5-image",
      "kind": "image",
      "label": "OpenAI: GPT-5 Image",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "GPT-5 Image combines OpenAI's GPT-5 model with state-of-the-art image generation capabilities. It offers major improvements in reasoning, code quality, and user experience while incorporating GPT Image 1's superior instruction following,...",
      "released": "2025-10-14",
      "price": "≈ $0.103 per image, on your AI tokens.",
      "priceShort": "≈ $0.103 / image"
    },
    {
      "id": "00/openai/gpt-5-image-mini",
      "orId": "openai/gpt-5-image-mini",
      "kind": "image",
      "label": "OpenAI: GPT-5 Image Mini",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "GPT-5 Image Mini combines OpenAI's advanced language capabilities, powered by GPT-5 Mini, with GPT Image 1 Mini for efficient image generation. This natively multimodal model features superior instruction following, text...",
      "released": "2025-10-16",
      "price": "≈ $0.021 per image, on your AI tokens.",
      "priceShort": "≈ $0.021 / image"
    },
    {
      "id": "00/openai/gpt-5.4-image-2",
      "orId": "openai/gpt-5.4-image-2",
      "kind": "image",
      "label": "OpenAI: GPT-5.4 Image 2",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "GPT-5.4 Image 2 combines OpenAI's GPT-5.4 model with state-of-the-art image generation capabilities from GPT Image 2. It enables rich multimodal workflows, allowing users to seamlessly move between reasoning, coding, and...",
      "released": "2026-04-21",
      "price": "≈ $0.077 per image, on your AI tokens.",
      "priceShort": "≈ $0.077 / image"
    },
    {
      "id": "00/openai/gpt-image-1",
      "orId": "openai/gpt-image-1",
      "kind": "image",
      "label": "OpenAI: GPT Image 1",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "OpenAI's GPT Image 1 generates and edits images via the dedicated Images API. Features accurate text rendering, transparent backgrounds, and up to 16 reference images for edits.",
      "released": "2026-06-24",
      "price": "≈ $0.103 per image, on your AI tokens.",
      "priceShort": "≈ $0.103 / image"
    },
    {
      "id": "00/openai/gpt-image-1-mini",
      "orId": "openai/gpt-image-1-mini",
      "kind": "image",
      "label": "OpenAI: GPT Image 1 Mini",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "A cost-efficient variant of GPT Image 1 for high-quality image generation at reduced latency and cost via OpenAI's dedicated Images API.",
      "released": "2026-06-24",
      "price": "≈ $0.021 per image, on your AI tokens.",
      "priceShort": "≈ $0.021 / image"
    },
    {
      "id": "00/openai/gpt-image-2",
      "orId": "openai/gpt-image-2",
      "kind": "image",
      "label": "OpenAI: GPT Image 2",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "OpenAI's latest image generation model. Supports high-fidelity image generation and editing via the dedicated Images API.",
      "released": "2026-06-24",
      "price": "≈ $0.077 per image, on your AI tokens.",
      "priceShort": "≈ $0.077 / image"
    },
    {
      "id": "00/openai/gpt-image-2.5-flare",
      "orId": "openai/gpt-image-2.5-flare",
      "kind": "image",
      "label": "OpenAI: GPT Image 2.5 Flare",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "GPT Image 2.5 Flare is an image generation and editing model from OpenAI, positioned as the speed-oriented tier of the GPT Image 2.5 series. It is suited to high-volume everyday...",
      "released": "2026-09-09",
      "price": "≈ $0.077 per image, on your AI tokens.",
      "priceShort": "≈ $0.077 / image"
    },
    {
      "id": "00/openai/gpt-image-2.5-sunburst",
      "orId": "openai/gpt-image-2.5-sunburst",
      "kind": "image",
      "label": "OpenAI: GPT Image 2.5 Sunburst",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "GPT Image 2.5 Sunburst is an image generation and editing model from OpenAI, positioned as the precision-oriented tier of the GPT Image 2.5 series. It is suited to detailed creative...",
      "released": "2026-09-09",
      "price": "≈ $0.077 per image, on your AI tokens.",
      "priceShort": "≈ $0.077 / image"
    },
    {
      "id": "00/openai/gpt-image-2.6-flare",
      "orId": "openai/gpt-image-2.6-flare",
      "kind": "image",
      "label": "OpenAI: GPT Image 2.6 Flare",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "GPT Image 2.6 Flare is OpenAI's image generation and editing model optimized for speed: fast iteration and interactive text-to-image and image-to-image workflows, generating about 50% faster than GPT Image 2.5....",
      "released": "2026-11-04",
      "price": "≈ $0.077 per image, on your AI tokens.",
      "priceShort": "≈ $0.077 / image"
    },
    {
      "id": "00/openai/gpt-image-2.6-sunburst",
      "orId": "openai/gpt-image-2.6-sunburst",
      "kind": "image",
      "label": "OpenAI: GPT Image 2.6 Sunburst",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "GPT Image 2.6 Sunburst is OpenAI's image generation and editing model optimized for detail: dense compositions, text-heavy layouts, and edits that keep details consistent from one edit to the next...",
      "released": "2026-11-04",
      "price": "≈ $0.077 per image, on your AI tokens.",
      "priceShort": "≈ $0.077 / image"
    },
    {
      "id": "00/qwen/qwen-image-2.1-pro",
      "orId": "qwen/qwen-image-2.1-pro",
      "kind": "image",
      "label": "Qwen: Qwen Image 2.1 Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Qwen Image 2.1 Pro is an image generation and editing model from Qwen, the higher-quality tier of the Qwen Image 2.1 series. Choose it over Qwen Image 2.1 Turbo when...",
      "released": "2026-10-16",
      "price": "$0.092 per image, on your AI tokens.",
      "priceShort": "$0.092 / image"
    },
    {
      "id": "00/qwen/qwen-image-2.1-turbo",
      "orId": "qwen/qwen-image-2.1-turbo",
      "kind": "image",
      "label": "Qwen: Qwen Image 2.1 Turbo",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Qwen Image 2.1 Turbo is the fast, lower-cost tier of the Qwen Image 2.1 series from Qwen, for image generation and editing where throughput and price matter more than peak...",
      "released": "2026-10-16",
      "price": "$0.037 per image, on your AI tokens.",
      "priceShort": "$0.037 / image"
    },
    {
      "id": "00/qwen/qwen-image-3",
      "orId": "qwen/qwen-image-3",
      "kind": "image",
      "label": "Qwen: Qwen Image 3",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Qwen Image 3 is a unified image generation and editing model from Qwen. It supports precise rendering of text and details as small as 10px, along with a richer world...",
      "released": "2026-08-05",
      "price": "$0.06 per image, on your AI tokens.",
      "priceShort": "$0.06 / image"
    },
    {
      "id": "00/qwen/qwen-image-3-pro",
      "orId": "qwen/qwen-image-3-pro",
      "kind": "image",
      "label": "Qwen: Qwen Image 3 Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Qwen Image 3 Pro is an image generation and editing model from Qwen. It supports precise rendering of text and details as small as 10px, along with richer world knowledge...",
      "released": "2026-08-05",
      "price": "$0.08 per image, on your AI tokens.",
      "priceShort": "$0.08 / image"
    },
    {
      "id": "00/recraft/recraft-v3",
      "orId": "recraft/recraft-v3",
      "kind": "image",
      "label": "Recraft: Recraft V3",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V3 is an image generation model from Recraft. It supports text and image inputs with image output at ~1K resolution across multiple aspect ratios. Supports the following `image_config` parameters:...",
      "released": "2026-05-07",
      "price": "$0.08 per image, on your AI tokens.",
      "priceShort": "$0.08 / image"
    },
    {
      "id": "00/recraft/recraft-v4",
      "orId": "recraft/recraft-v4",
      "kind": "image",
      "label": "Recraft: Recraft V4",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V4 is an image generation model from Recraft. It supports text and image inputs with image output at ~1K resolution across multiple aspect ratios. It delivers stronger compositional judgment,...",
      "released": "2026-05-07",
      "price": "$0.08 per image, on your AI tokens.",
      "priceShort": "$0.08 / image"
    },
    {
      "id": "00/recraft/recraft-v4-pro",
      "orId": "recraft/recraft-v4-pro",
      "kind": "image",
      "label": "Recraft: Recraft V4 Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V4 Pro is an image generation model from Recraft. It supports text and image inputs with image output at ~2K resolution across multiple aspect ratios, double the resolution of...",
      "released": "2026-05-07",
      "price": "$0.5 per image, on your AI tokens.",
      "priceShort": "$0.5 / image"
    },
    {
      "id": "00/recraft/recraft-v4-pro-vector",
      "orId": "recraft/recraft-v4-pro-vector",
      "kind": "image",
      "label": "Recraft: Recraft V4 Pro Vector",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V4 Pro Vector is the vector (SVG) variant of Recraft V4 Pro. It supports text and image inputs and produces vector image output across multiple aspect ratios at the...",
      "released": "2026-05-13",
      "price": "$0.6 per image, on your AI tokens.",
      "priceShort": "$0.6 / image"
    },
    {
      "id": "00/recraft/recraft-v4-styles",
      "orId": "recraft/recraft-v4-styles",
      "kind": "image",
      "label": "Recraft: Recraft V4 Styles",
      "modes": [
        "image→image"
      ],
      "kinds": [
        "image-edit"
      ],
      "description": "Recraft V4 Styles is a style-consistent image generation model from Recraft. Every request requires at least one style reference image and generates a new image that reproduces the reference's rendering...",
      "released": "2026-08-26",
      "price": "$0.07 per image, on your AI tokens.",
      "priceShort": "$0.07 / image"
    },
    {
      "id": "00/recraft/recraft-v4-styles-pro",
      "orId": "recraft/recraft-v4-styles-pro",
      "kind": "image",
      "label": "Recraft: Recraft V4 Styles Pro",
      "modes": [
        "image→image"
      ],
      "kinds": [
        "image-edit"
      ],
      "description": "Recraft V4 Styles Pro is a style-consistent image generation model from Recraft. Every request requires at least one style reference image and generates a new image that reproduces the reference's...",
      "released": "2026-08-26",
      "price": "$0.2 per image, on your AI tokens.",
      "priceShort": "$0.2 / image"
    },
    {
      "id": "00/recraft/recraft-v4-styles-pro-vector",
      "orId": "recraft/recraft-v4-styles-pro-vector",
      "kind": "image",
      "label": "Recraft: Recraft V4 Styles Pro Vector",
      "modes": [
        "image→image"
      ],
      "kinds": [
        "image-edit"
      ],
      "description": "Recraft V4 Styles Pro Vector is a style-consistent image generation model from Recraft. Every request requires at least one style reference image and generates a new image that reproduces the...",
      "released": "2026-08-26",
      "price": "$0.24 per image, on your AI tokens.",
      "priceShort": "$0.24 / image"
    },
    {
      "id": "00/recraft/recraft-v4-styles-vector",
      "orId": "recraft/recraft-v4-styles-vector",
      "kind": "image",
      "label": "Recraft: Recraft V4 Styles Vector",
      "modes": [
        "image→image"
      ],
      "kinds": [
        "image-edit"
      ],
      "description": "Recraft V4 Styles Vector is a style-consistent image generation model from Recraft. Every request requires at least one style reference image and generates a new image that reproduces the reference's...",
      "released": "2026-08-26",
      "price": "$0.1 per image, on your AI tokens.",
      "priceShort": "$0.1 / image"
    },
    {
      "id": "00/recraft/recraft-v4-vector",
      "orId": "recraft/recraft-v4-vector",
      "kind": "image",
      "label": "Recraft: Recraft V4 Vector",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V4 Vector is the vector (SVG) variant of Recraft V4. It supports text and image inputs and produces vector image output across multiple aspect ratios. Compared to the raster...",
      "released": "2026-05-13",
      "price": "$0.16 per image, on your AI tokens.",
      "priceShort": "$0.16 / image"
    },
    {
      "id": "00/recraft/recraft-v4.1",
      "orId": "recraft/recraft-v4.1",
      "kind": "image",
      "label": "Recraft: Recraft V4.1",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V4.1 is an image generation model from Recraft tuned for high aesthetics. It supports text and image inputs with image output at ~1K resolution across multiple aspect ratios, with...",
      "released": "2026-05-13",
      "price": "$0.07 per image, on your AI tokens.",
      "priceShort": "$0.07 / image"
    },
    {
      "id": "00/recraft/recraft-v4.1-flash",
      "orId": "recraft/recraft-v4.1-flash",
      "kind": "image",
      "label": "Recraft: Recraft V4.1 Flash",
      "modes": [
        "text→image"
      ],
      "kinds": [
        "image"
      ],
      "description": "Recraft V4.1 Flash is a text-to-image model from Recraft, the speed and cost tier of the V4.1 family. It generates ~1K raster images in about 1.5 seconds end to end,...",
      "released": "2026-09-23",
      "price": "$0.014 per image, on your AI tokens.",
      "priceShort": "$0.014 / image"
    },
    {
      "id": "00/recraft/recraft-v4.1-flash-v2",
      "orId": "recraft/recraft-v4.1-flash-v2",
      "kind": "image",
      "label": "Recraft: Recraft V4.1 Flash V2",
      "modes": [
        "text→image"
      ],
      "kinds": [
        "image"
      ],
      "description": "Recraft V4.1 Flash V2 is a text-to-image model from Recraft, an updated version of [Recraft V4.1 Flash](/recraft/recraft-v4.1-flash) in the speed tier of the V4.1 family. It is suited for rapid...",
      "released": "2026-11-13",
      "price": "$0.014 per image, on your AI tokens.",
      "priceShort": "$0.014 / image"
    },
    {
      "id": "00/recraft/recraft-v4.1-pro",
      "orId": "recraft/recraft-v4.1-pro",
      "kind": "image",
      "label": "Recraft: Recraft V4.1 Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V4.1 Pro is an image generation model from Recraft tuned for high aesthetics. It supports text and image inputs with image output at ~2K resolution across multiple aspect ratios...",
      "released": "2026-05-13",
      "price": "$0.42 per image, on your AI tokens.",
      "priceShort": "$0.42 / image"
    },
    {
      "id": "00/recraft/recraft-v4.1-pro-vector",
      "orId": "recraft/recraft-v4.1-pro-vector",
      "kind": "image",
      "label": "Recraft: Recraft V4.1 Pro Vector",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V4.1 Pro Vector is the vector (SVG) variant of Recraft V4.1 Pro, tuned for high aesthetics. It supports text and image inputs and produces higher-resolution SVG image output across...",
      "released": "2026-05-13",
      "price": "$0.6 per image, on your AI tokens.",
      "priceShort": "$0.6 / image"
    },
    {
      "id": "00/recraft/recraft-v4.1-utility",
      "orId": "recraft/recraft-v4.1-utility",
      "kind": "image",
      "label": "Recraft: Recraft V4.1 Utility",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V4.1 Utility is a general-purpose image generation model from Recraft. It supports text and image inputs with image output at ~1K resolution across multiple aspect ratios, with typical generation...",
      "released": "2026-05-13",
      "price": "$0.07 per image, on your AI tokens.",
      "priceShort": "$0.07 / image"
    },
    {
      "id": "00/recraft/recraft-v4.1-utility-pro",
      "orId": "recraft/recraft-v4.1-utility-pro",
      "kind": "image",
      "label": "Recraft: Recraft V4.1 Utility Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V4.1 Utility Pro is a general-purpose image generation model from Recraft. It supports text and image inputs with image output at ~2K resolution across multiple aspect ratios — double...",
      "released": "2026-05-13",
      "price": "$0.42 per image, on your AI tokens.",
      "priceShort": "$0.42 / image"
    },
    {
      "id": "00/recraft/recraft-v4.1-vector",
      "orId": "recraft/recraft-v4.1-vector",
      "kind": "image",
      "label": "Recraft: Recraft V4.1 Vector",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Recraft V4.1 Vector is the vector (SVG) variant of Recraft V4.1, tuned for high aesthetics. It supports text and image inputs and produces SVG image output across multiple aspect ratios,...",
      "released": "2026-05-13",
      "price": "$0.16 per image, on your AI tokens.",
      "priceShort": "$0.16 / image"
    },
    {
      "id": "00/runway/aleph-2",
      "orId": "runway/aleph-2",
      "kind": "video",
      "label": "Runway: Aleph 2.0",
      "modes": [
        "video→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Runway Aleph 2.0 is an in-context video editing model from Runway. It applies text instructions and keyframe-guided edits across existing footage while preserving details that are not meant to change....",
      "released": "2026-07-29",
      "price": "$0.56 per second of video, on your AI tokens.",
      "priceShort": "$0.56 / s"
    },
    {
      "id": "00/runway/gen-4.5",
      "orId": "runway/gen-4.5",
      "kind": "video",
      "label": "Runway: Gen-4.5",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Runway Gen-4.5 is a video generation model from Runway for text-to-video and image-to-video workflows. It is designed for cinematic scene creation with strong motion quality, visual fidelity, and prompt adherence....",
      "released": "2026-07-29",
      "durations": [
        2,
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10
      ],
      "resolutions": [
        "720p"
      ],
      "price": "$0.24 per second of video, on your AI tokens.",
      "priceShort": "$0.24 / s"
    },
    {
      "id": "00/sourceful/riverflow-v2-fast",
      "orId": "sourceful/riverflow-v2-fast",
      "kind": "image",
      "label": "Sourceful: Riverflow V2 Fast",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Riverflow V2 Fast is the fastest variant of Sourceful's Riverflow 2.0 lineup, best for production deployments and latency-critical workflows. The Riverflow 2.0 series represents SOTA performance on image generation and...",
      "released": "2026-02-02",
      "price": "$0.04 per image, on your AI tokens.",
      "priceShort": "$0.04 / image"
    },
    {
      "id": "00/sourceful/riverflow-v2-pro",
      "orId": "sourceful/riverflow-v2-pro",
      "kind": "image",
      "label": "Sourceful: Riverflow V2 Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Riverflow V2 Pro is the most powerful variant of Sourceful's Riverflow 2.0 lineup, best for top-tier control and perfect text rendering. The Riverflow 2.0 series represents SOTA performance on image...",
      "released": "2026-02-02",
      "price": "$0.3 per image, on your AI tokens.",
      "priceShort": "$0.3 / image"
    },
    {
      "id": "00/sourceful/riverflow-v2.5-fast",
      "orId": "sourceful/riverflow-v2.5-fast",
      "kind": "image",
      "label": "Sourceful: Riverflow V2.5 Fast",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Riverflow V2.5 Fast is the speed-optimized variant of Sourceful's Riverflow 2.5 lineup, best for production deployments and latency-critical workflows. The Riverflow 2.5 series is a unified text-to-image and image-to-image family...",
      "released": "2026-06-04",
      "price": "$0.038 per image, on your AI tokens.",
      "priceShort": "$0.038 / image"
    },
    {
      "id": "00/sourceful/riverflow-v2.5-pro",
      "orId": "sourceful/riverflow-v2.5-pro",
      "kind": "image",
      "label": "Sourceful: Riverflow V2.5 Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Riverflow V2.5 Pro is the most powerful variant of Sourceful's Riverflow 2.5 lineup, best for top-tier control and quality-sensitive outputs. The Riverflow 2.5 series is a unified text-to-image and image-to-image...",
      "released": "2026-06-04",
      "price": "$0.26 per image, on your AI tokens.",
      "priceShort": "$0.26 / image"
    },
    {
      "id": "00/sourceful/riverflow-v2.6-pro",
      "orId": "sourceful/riverflow-v2.6-pro",
      "kind": "image",
      "label": "Sourceful: Riverflow V2.6 Pro",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Riverflow 2.6 Pro is the flagship model of Sourceful's Riverflow 2.6 lineup, a unified text-to-image and image-to-image family that treats generation as a production workflow: an integrated reasoning model plans...",
      "released": "2026-10-29",
      "price": "$0.26 per image, on your AI tokens.",
      "priceShort": "$0.26 / image"
    },
    {
      "id": "00/tbc/mischief-1",
      "orId": "tbc/mischief-1",
      "kind": "video",
      "label": "TBC: mischief 1",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "TBC mischief 1 is the world's first neuron-derived video model from The Biological Computing Co. It supports image to video and is a fast and affordable way to bring ideas...",
      "released": "2026-10-30",
      "durations": [
        3,
        5
      ],
      "resolutions": [
        "480p",
        "720p"
      ],
      "price": "$0.01 per second of video at 480p, up to $0.02/s at 720p, on your AI tokens.",
      "priceShort": "$0.01 / s (480p)"
    },
    {
      "id": "00/tencent/hy-image-v3.5-preview",
      "orId": "tencent/hy-image-v3.5-preview",
      "kind": "image",
      "label": "Tencent: Hy Image 3.5 Preview",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Hy Image 3.5 Preview is a unified image generation and editing model from Tencent. It handles text-to-image, image-to-image, and multi-turn editing through one endpoint, taking up to 20 reference images...",
      "released": "2026-10-05",
      "price": "≈ $0.004 per image, on your AI tokens.",
      "priceShort": "≈ $0.004 / image"
    },
    {
      "id": "00/tencent/hy-world-2.1-panorama",
      "orId": "tencent/hy-world-2.1-panorama",
      "kind": "image",
      "label": "Tencent: HY-World 2.1 Panorama",
      "modes": [
        "text→image"
      ],
      "kinds": [
        "image"
      ],
      "description": "HY-World 2.1 Panorama is a 360-degree panorama generation model from Tencent, part of the Hunyuan World family. It turns a single text prompt into one high-resolution equirectangular panorama suitable for...",
      "released": "2026-11-13",
      "price": "≈ $0.004 per image, on your AI tokens.",
      "priceShort": "≈ $0.004 / image"
    },
    {
      "id": "00/x-ai/grok-imagine-image-2.0",
      "orId": "x-ai/grok-imagine-image-2.0",
      "kind": "image",
      "label": "xAI: Grok Imagine Image 2.0",
      "modes": [
        "text→image",
        "image→image"
      ],
      "kinds": [
        "image",
        "image-edit"
      ],
      "description": "Grok Imagine Image 2.0 is an image generation and editing model from xAI. It is suited for creating images from text prompts and editing images from references, with low and...",
      "released": "2026-08-11",
      "price": "$0.08 per image, on your AI tokens.",
      "priceShort": "$0.08 / image"
    },
    {
      "id": "00/x-ai/grok-imagine-video",
      "orId": "x-ai/grok-imagine-video",
      "kind": "video",
      "label": "SpaceXAI: Grok Imagine Video",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Grok Imagine Video is SpaceXAI's fast, text-, image-, and reference-conditioned video generation model. It produces short videos (1–15 seconds, 24 fps) at 480p or 720p across seven aspect ratios -...",
      "released": "2026-05-18",
      "durations": [
        1,
        2,
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "480p",
        "720p"
      ],
      "price": "$0.1 per second of video at 480p, up to $0.14/s at 720p, on your AI tokens.",
      "priceShort": "$0.1 / s (480p)"
    },
    {
      "id": "00/x-ai/grok-imagine-video-1.5",
      "orId": "x-ai/grok-imagine-video-1.5",
      "kind": "video",
      "label": "SpaceXAI: Grok Imagine Video 1.5",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Grok Imagine Video 1.5 is a video generation model from SpaceXAI. It creates videos from text prompts, with an optional starting image to guide the scene. It can direct subject...",
      "released": "2026-07-20",
      "durations": [
        1,
        2,
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "480p",
        "720p",
        "1080p"
      ],
      "price": "$0.16 per second of video at 480p, up to $0.5/s at 1080p, on your AI tokens.",
      "priceShort": "$0.16 / s (480p)"
    },
    {
      "id": "00/x-ai/grok-imagine-video-1.5-lite",
      "orId": "x-ai/grok-imagine-video-1.5-lite",
      "kind": "video",
      "label": "SpaceXAI: Grok Imagine Video 1.5 Lite",
      "modes": [
        "text→video",
        "image→video"
      ],
      "kinds": [
        "video"
      ],
      "description": "Grok Imagine Video 1.5 Lite is a faster, lower-cost video generation model from SpaceXAI, distilled from [Grok Imagine Video 1.5](/x-ai/grok-imagine-video-1.5). It supports text-to-video and image-to-video, trading some quality for speed...",
      "released": "2026-10-06",
      "durations": [
        1,
        2,
        3,
        4,
        5,
        6,
        7,
        8,
        9,
        10,
        11,
        12,
        13,
        14,
        15
      ],
      "resolutions": [
        "480p",
        "720p",
        "1080p"
      ],
      "price": "$0.04 per second of video at 480p, up to $0.28/s at 1080p, on your AI tokens.",
      "priceShort": "$0.04 / s (480p)"
    }
  ],
  "falFetchedAt": "2026-11-15T06:40:27.471Z",
  "falScanned": 1432,
  "pricesChecked": "2026-07-29",
  "charsPerMinute": 900,
  "samplesBase": "https://dl.0-0.chat/media-samples",
  "publishedAt": "2026-11-16T06:40:06.316Z"
}