← All comparisons
Devstral 2 vs Claude 4.1 Opus (Non-reasoning)
Mistral vs Anthropic — side-by-side benchmark comparison
| Devstral 2 | Claude 4.1 Opus (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 22.0 | 36.0 |
| Coding Index | 23.7 | — |
| Math Index | 36.7 | — |
| Output speed (tok/s) | 79.9 | 44.7 |
| Blended price ($/1M) | $0.00 | $32.81 |
| Time to first token (s) | 0.50s | 1.63s |
| aime | — | — |
| aime 25 | 36.7% | — |
| artificial analysis coding index | 23.70 | — |
| artificial analysis intelligence index | 22.00 | 36.00 |
| artificial analysis math index | 36.70 | — |
| gpqa | 59.4% | — |
| hle | 3.6% | — |
| ifbench | 38.1% | — |
| lcr | 30.0% | — |
| livecodebench | 44.8% | — |
| math 500 | — | — |
| mmlu pro | 76.2% | — |
| scicode | 33.1% | — |
| tau2 | 24.9% | — |
| terminalbench hard | 18.9% | — |
Benchmark data from Artificial Analysis.