← All comparisons
Claude 4.5 Haiku (Non-reasoning) vs Hermes 4 - Llama-3.1 70B (Reasoning)
Anthropic vs Nous Research — side-by-side benchmark comparison
| Claude 4.5 Haiku (Non-reasoning) | Hermes 4 - Llama-3.1 70B (Reasoning) | |
|---|---|---|
| Intelligence Index | 31.0 | 16.0 |
| Coding Index | 29.6 | 14.4 |
| Math Index | 39.0 | 68.7 |
| Output speed (tok/s) | 105.8 | 92.8 |
| Blended price ($/1M) | $2.19 | $0.20 |
| Time to first token (s) | 0.65s | 0.64s |
| aime | — | — |
| aime 25 | 39.0% | 68.7% |
| artificial analysis coding index | 29.60 | 14.40 |
| artificial analysis intelligence index | 31.00 | 16.00 |
| artificial analysis math index | 39.00 | 68.70 |
| gpqa | 64.6% | 69.9% |
| hle | 4.3% | 7.9% |
| ifbench | 42.0% | 31.3% |
| lcr | 43.7% | 6.7% |
| livecodebench | 51.1% | 65.3% |
| math 500 | — | — |
| mmlu pro | 80.0% | 81.1% |
| scicode | 34.4% | 34.1% |
| tau2 | 32.5% | 22.5% |
| terminalbench hard | 27.3% | 4.5% |
Benchmark data from Artificial Analysis.