← All comparisons
Hermes 4 - Llama-3.1 70B (Reasoning) vs Qwen3.5 27B (Reasoning)
Nous Research vs Alibaba — side-by-side benchmark comparison
| Hermes 4 - Llama-3.1 70B (Reasoning) | Qwen3.5 27B (Reasoning) | |
|---|---|---|
| Intelligence Index | 16.0 | 42.1 |
| Coding Index | 14.4 | 34.9 |
| Math Index | 68.7 | — |
| Output speed (tok/s) | 92.8 | 84.3 |
| Blended price ($/1M) | $0.20 | $0.82 |
| Time to first token (s) | 0.64s | 1.37s |
| aime | — | — |
| aime 25 | 68.7% | — |
| artificial analysis coding index | 14.40 | 34.90 |
| artificial analysis intelligence index | 16.00 | 42.10 |
| artificial analysis math index | 68.70 | — |
| gpqa | 69.9% | 85.8% |
| hle | 7.9% | 22.2% |
| ifbench | 31.3% | 75.6% |
| lcr | 6.7% | 67.3% |
| livecodebench | 65.3% | — |
| math 500 | — | — |
| mmlu pro | 81.1% | — |
| scicode | 34.1% | 39.5% |
| tau2 | 22.5% | 93.9% |
| terminalbench hard | 4.5% | 32.6% |
Benchmark data from Artificial Analysis.