← All comparisons
Nemotron 3 Nano Omni 30B A3B Reasoning vs DeepSeek R1 Distill Llama 70B
NVIDIA vs DeepSeek — side-by-side benchmark comparison
| Nemotron 3 Nano Omni 30B A3B Reasoning | DeepSeek R1 Distill Llama 70B | |
|---|---|---|
| Intelligence Index | 21.4 | 16.0 |
| Coding Index | 14.8 | 11.4 |
| Math Index | — | 53.7 |
| Output speed (tok/s) | 285.9 | 46.8 |
| Blended price ($/1M) | $0.13 | $0.79 |
| Time to first token (s) | 0.58s | 0.33s |
| aime | — | 67.0% |
| aime 25 | — | 53.7% |
| artificial analysis coding index | 14.80 | 11.40 |
| artificial analysis intelligence index | 21.40 | 16.00 |
| artificial analysis math index | — | 53.70 |
| gpqa | 46.9% | 40.2% |
| hle | 5.3% | 6.1% |
| ifbench | 63.2% | 27.6% |
| lcr | 35.7% | 11.0% |
| livecodebench | — | 26.6% |
| math 500 | — | 93.5% |
| mmlu pro | — | 79.5% |
| scicode | 27.8% | 31.3% |
| tau2 | 45.3% | 21.9% |
| terminalbench hard | 8.3% | 1.5% |
Benchmark data from Artificial Analysis.