← All comparisons
Devstral Small 2 vs Qwen3 235B A22B 2507 Instruct
Mistral vs Alibaba — side-by-side benchmark comparison
| Devstral Small 2 | Qwen3 235B A22B 2507 Instruct | |
|---|---|---|
| Intelligence Index | 19.5 | 25.0 |
| Coding Index | 20.7 | 22.1 |
| Math Index | 34.3 | 71.7 |
| Output speed (tok/s) | 75.2 | 57.0 |
| Blended price ($/1M) | $0.00 | $0.36 |
| Time to first token (s) | 0.73s | 1.34s |
| aime | — | 71.7% |
| aime 25 | 34.3% | 71.7% |
| artificial analysis coding index | 20.70 | 22.10 |
| artificial analysis intelligence index | 19.50 | 25.00 |
| artificial analysis math index | 34.30 | 71.70 |
| gpqa | 53.2% | 75.3% |
| hle | 3.4% | 10.6% |
| ifbench | 31.2% | 46.1% |
| lcr | 24.0% | 31.2% |
| livecodebench | 34.8% | 52.4% |
| math 500 | — | 98.0% |
| mmlu pro | 67.8% | 82.8% |
| scicode | 28.8% | 36.0% |
| tau2 | 23.4% | 33.3% |
| terminalbench hard | 16.7% | 15.2% |
Benchmark data from Artificial Analysis.