← All comparisons
Magistral Medium 1.2 vs Hermes 4 - Llama-3.1 405B (Non-reasoning)
Mistral vs Nous Research — side-by-side benchmark comparison
| Magistral Medium 1.2 | Hermes 4 - Llama-3.1 405B (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 27.1 | 17.6 |
| Coding Index | 21.7 | 18.1 |
| Math Index | 82.0 | 15.3 |
| Output speed (tok/s) | 39.4 | 40.8 |
| Blended price ($/1M) | $2.75 | $1.50 |
| Time to first token (s) | 0.54s | 0.73s |
| aime | — | — |
| aime 25 | 82.0% | 15.3% |
| artificial analysis coding index | 21.70 | 18.10 |
| artificial analysis intelligence index | 27.10 | 17.60 |
| artificial analysis math index | 82.00 | 15.30 |
| gpqa | 73.9% | 53.6% |
| hle | 9.6% | 4.2% |
| ifbench | 43.0% | 34.8% |
| lcr | 51.3% | 20.0% |
| livecodebench | 75.0% | 54.6% |
| math 500 | — | — |
| mmlu pro | 81.5% | 72.9% |
| scicode | 39.2% | 34.6% |
| tau2 | 52.0% | 26.6% |
| terminalbench hard | 12.9% | 9.8% |
Benchmark data from Artificial Analysis.