← All comparisons
Llama Nemotron Super 49B v1.5 (Non-reasoning) vs DeepSeek R1 Distill Llama 70B
NVIDIA vs DeepSeek — side-by-side benchmark comparison
| Llama Nemotron Super 49B v1.5 (Non-reasoning) | DeepSeek R1 Distill Llama 70B | |
|---|---|---|
| Intelligence Index | 14.6 | 16.0 |
| Coding Index | 10.5 | 11.4 |
| Math Index | 8.0 | 53.7 |
| Output speed (tok/s) | 48.9 | 46.8 |
| Blended price ($/1M) | $0.17 | $0.79 |
| Time to first token (s) | 0.28s | 0.33s |
| aime | 13.7% | 67.0% |
| aime 25 | 8.0% | 53.7% |
| artificial analysis coding index | 10.50 | 11.40 |
| artificial analysis intelligence index | 14.60 | 16.00 |
| artificial analysis math index | 8.00 | 53.70 |
| gpqa | 48.1% | 40.2% |
| hle | 4.3% | 6.1% |
| ifbench | 32.9% | 27.6% |
| lcr | 22.0% | 11.0% |
| livecodebench | 29.0% | 26.6% |
| math 500 | 77.0% | 93.5% |
| mmlu pro | 69.2% | 79.5% |
| scicode | 23.8% | 31.3% |
| tau2 | 25.1% | 21.9% |
| terminalbench hard | 3.8% | 1.5% |
Benchmark data from Artificial Analysis.