← All comparisons
Claude 4.1 Opus (Reasoning) vs Devstral Small (May '25)
Anthropic vs Mistral — side-by-side benchmark comparison
| Claude 4.1 Opus (Reasoning) | Devstral Small (May '25) | |
|---|---|---|
| Intelligence Index | 42.0 | 18.0 |
| Coding Index | 36.5 | 12.2 |
| Math Index | 80.3 | — |
| Output speed (tok/s) | 44.5 | 0.0 |
| Blended price ($/1M) | $32.81 | $0.00 |
| Time to first token (s) | 8.55s | 0.00s |
| aime | — | 6.7% |
| aime 25 | 80.3% | — |
| artificial analysis coding index | 36.50 | 12.20 |
| artificial analysis intelligence index | 42.00 | 18.00 |
| artificial analysis math index | 80.30 | — |
| gpqa | 80.9% | 43.4% |
| hle | 11.9% | 4.0% |
| ifbench | 55.4% | 31.6% |
| lcr | 66.3% | 26.7% |
| livecodebench | 65.4% | 25.8% |
| math 500 | — | 68.4% |
| mmlu pro | 88.0% | 63.2% |
| scicode | 40.9% | 24.5% |
| tau2 | 71.4% | 38.0% |
| terminalbench hard | 34.3% | 6.1% |
Benchmark data from Artificial Analysis.