← All comparisons
Nemotron 3 Nano Omni 30B A3B Reasoning vs Hermes 4 - Llama-3.1 70B (Reasoning)
NVIDIA vs Nous Research — side-by-side benchmark comparison
| Nemotron 3 Nano Omni 30B A3B Reasoning | Hermes 4 - Llama-3.1 70B (Reasoning) | |
|---|---|---|
| Intelligence Index | 21.4 | 16.0 |
| Coding Index | 14.8 | 14.4 |
| Math Index | — | 68.7 |
| Output speed (tok/s) | 285.9 | 92.8 |
| Blended price ($/1M) | $0.13 | $0.20 |
| Time to first token (s) | 0.58s | 0.64s |
| aime | — | — |
| aime 25 | — | 68.7% |
| artificial analysis coding index | 14.80 | 14.40 |
| artificial analysis intelligence index | 21.40 | 16.00 |
| artificial analysis math index | — | 68.70 |
| gpqa | 46.9% | 69.9% |
| hle | 5.3% | 7.9% |
| ifbench | 63.2% | 31.3% |
| lcr | 35.7% | 6.7% |
| livecodebench | — | 65.3% |
| math 500 | — | — |
| mmlu pro | — | 81.1% |
| scicode | 27.8% | 34.1% |
| tau2 | 45.3% | 22.5% |
| terminalbench hard | 8.3% | 4.5% |
Benchmark data from Artificial Analysis.