← All comparisons
GPT-5 nano (high) vs DeepSeek R1 Distill Llama 8B
OpenAI vs DeepSeek — side-by-side benchmark comparison
| GPT-5 nano (high) | DeepSeek R1 Distill Llama 8B | |
|---|---|---|
| Intelligence Index | 26.8 | 12.1 |
| Coding Index | 20.3 | — |
| Math Index | 83.7 | 41.3 |
| Output speed (tok/s) | 173.3 | 0.0 |
| Blended price ($/1M) | $0.14 | $0.00 |
| Time to first token (s) | 78.29s | 0.00s |
| aime | — | 33.3% |
| aime 25 | 83.7% | 41.3% |
| artificial analysis coding index | 20.30 | — |
| artificial analysis intelligence index | 26.80 | 12.10 |
| artificial analysis math index | 83.70 | 41.30 |
| gpqa | 67.6% | 30.2% |
| hle | 8.2% | 4.2% |
| ifbench | 67.6% | 17.6% |
| lcr | 41.7% | 0.0% |
| livecodebench | 78.9% | 23.3% |
| math 500 | — | 85.3% |
| mmlu pro | 78.0% | 54.3% |
| scicode | 36.6% | 11.9% |
| tau2 | 36.5% | — |
| terminalbench hard | 12.1% | — |
Benchmark data from Artificial Analysis.