DeepSeek: DeepSeek V3
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...
Everything else
| Max output | 16K |
|---|---|
| Input / 1M | $0.257 |
| Output / 1M | $1.03 |
| Cache read / 1M | — |
| Cache write / 1M | — |
| 1M in + 1M out | $1.29 |
| Takes | text |
| Returns | text |
| Tools | yes |
| Reasoning | no |
| Open weights | yes |
| Released | 2024-12-26 |
| Knowledge cutoff | 2024-07-31 |
| Retires | — |
| Catalog id | deepseek/deepseek-chat |
Benchmarks
| source | epoch |
|---|---|
| eci | 133 |
| confidence | Confident |
| accessibility | Open weights (restricted use) |
| gpqa | 0.712 |
| gpqaStderr | 0.0323 |
| aime | 0.489 |
| aimeStderr | 0.0754 |
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.