Compare models
Pick up to three models.
 Mistral Large 3 |  DeepSeek V4 Pro |  Qwen3 235B | |
|---|---|---|---|
| AA Intelligence Index | 16 | 44 | 20 |
| Context window | 128K tokens | 1M tokens | 256K tokens |
| Input $/MTok | $2 / MTok | $0.435 / MTok | — |
| Output $/MTok | $6 / MTok | $0.87 / MTok | — |
| Modalities | text, image, code | text, code | text, code |
| API | Yes | Yes | Yes |
| Open source | Yes | Yes | Yes |
Intelligence Index scores are sourced from Artificial Analysis, an independent evaluator — not Timace's own judgment.