← All comparisons
gpt-oss-20B (high) vs Llama Nemotron Super 49B v1.5 (Reasoning)
OpenAI vs NVIDIA — side-by-side benchmark comparison
| gpt-oss-20B (high) | Llama Nemotron Super 49B v1.5 (Reasoning) | |
|---|---|---|
| Intelligence Index | 24.5 | 18.7 |
| Coding Index | 18.5 | 15.1 |
| Math Index | 89.3 | 76.7 |
| Output speed (tok/s) | 267.0 | 48.1 |
| Blended price ($/1M) | $0.09 | $0.17 |
| Time to first token (s) | 0.42s | 0.30s |
| aime | — | 86.0% |
| aime 25 | 89.3% | 76.7% |
| artificial analysis coding index | 18.50 | 15.10 |
| artificial analysis intelligence index | 24.50 | 18.70 |
| artificial analysis math index | 89.30 | 76.70 |
| gpqa | 68.8% | 74.8% |
| hle | 9.8% | 6.8% |
| ifbench | 65.1% | 37.0% |
| lcr | 30.7% | 34.0% |
| livecodebench | 77.7% | 73.7% |
| math 500 | — | 98.3% |
| mmlu pro | 74.8% | 81.4% |
| scicode | 34.4% | 34.8% |
| tau2 | 60.2% | 28.1% |
| terminalbench hard | 10.6% | 5.3% |
Benchmark data from Artificial Analysis.