← All comparisons
Hermes 4 - Llama-3.1 70B (Non-reasoning) vs Claude 4.1 Opus (Reasoning)
Nous Research vs Anthropic — side-by-side benchmark comparison
| Hermes 4 - Llama-3.1 70B (Non-reasoning) | Claude 4.1 Opus (Reasoning) | |
|---|---|---|
| Intelligence Index | 12.6 | 42.0 |
| Coding Index | 9.2 | 36.5 |
| Math Index | 11.3 | 80.3 |
| Output speed (tok/s) | 94.3 | 44.5 |
| Blended price ($/1M) | $0.20 | $32.81 |
| Time to first token (s) | 0.61s | 8.55s |
| aime | — | — |
| aime 25 | 11.3% | 80.3% |
| artificial analysis coding index | 9.20 | 36.50 |
| artificial analysis intelligence index | 12.60 | 42.00 |
| artificial analysis math index | 11.30 | 80.30 |
| gpqa | 49.1% | 80.9% |
| hle | 3.6% | 11.9% |
| ifbench | 29.0% | 55.4% |
| lcr | 2.0% | 66.3% |
| livecodebench | 26.9% | 65.4% |
| math 500 | — | — |
| mmlu pro | 66.4% | 88.0% |
| scicode | 27.7% | 40.9% |
| tau2 | 21.6% | 71.4% |
| terminalbench hard | 0.0% | 34.3% |
Benchmark data from Artificial Analysis.