← All comparisons
Magistral Medium 1.2 vs GPT-5 nano (minimal)
Mistral vs OpenAI — side-by-side benchmark comparison
| Magistral Medium 1.2 | GPT-5 nano (minimal) | |
|---|---|---|
| Intelligence Index | 27.1 | 13.8 |
| Coding Index | 21.7 | 14.2 |
| Math Index | 82.0 | 27.3 |
| Output speed (tok/s) | 39.4 | 175.6 |
| Blended price ($/1M) | $2.75 | $0.14 |
| Time to first token (s) | 0.54s | 0.67s |
| aime | — | — |
| aime 25 | 82.0% | 27.3% |
| artificial analysis coding index | 21.70 | 14.20 |
| artificial analysis intelligence index | 27.10 | 13.80 |
| artificial analysis math index | 82.00 | 27.30 |
| gpqa | 73.9% | 42.8% |
| hle | 9.6% | 4.1% |
| ifbench | 43.0% | 32.5% |
| lcr | 51.3% | 20.0% |
| livecodebench | 75.0% | 47.0% |
| math 500 | — | — |
| mmlu pro | 81.5% | 55.6% |
| scicode | 39.2% | 29.1% |
| tau2 | 52.0% | 25.7% |
| terminalbench hard | 12.9% | 6.8% |
Benchmark data from Artificial Analysis.