MoonshotAI: Kimi K3
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...
Everything else
| Max output | — |
|---|---|
| Input / 1M | $3 |
| Output / 1M | $15 |
| Cache read / 1M | $0.3 |
| Cache write / 1M | — |
| 1M in + 1M out | $18 |
| Takes | text, image |
| Returns | text |
| Tools | yes |
| Reasoning | optional · max, high, low |
| Open weights | yes |
| Released | 2026-07-16 |
| Knowledge cutoff | — |
| Retires | — |
| Catalog id | moonshotai/kimi-k3 |
Benchmarks
| source | epoch |
|---|---|
| eci | 155.59 |
| confidence | Speculative |
| accessibility | Open weights (non-commercial) |
| gpqa | 0.931 |
| gpqaStderr | 0.0149 |
| aime | 0.972 |
| aimeStderr | 0.0112 |
| simpleQA | 0.427 |
| simpleQAStderr | 0.0156 |
Benchmark scores by Epoch AI, CC BY 4.0.
Using it
In 00, this model is picked per agent — and per tier, so the model that answers your customers need not be the one that writes your code. Get 00.