← All comparisons
DeepSeek R1 Distill Llama 70B vs DeepSeek R1 Distill Qwen 1.5B
DeepSeek vs DeepSeek — side-by-side benchmark comparison
| DeepSeek R1 Distill Llama 70B | DeepSeek R1 Distill Qwen 1.5B | |
|---|---|---|
| Intelligence Index | 16.0 | 9.1 |
| Coding Index | 11.4 | — |
| Math Index | 53.7 | 22.0 |
| Output speed (tok/s) | 46.8 | 0.0 |
| Blended price ($/1M) | $0.79 | $0.00 |
| Time to first token (s) | 0.33s | 0.00s |
| aime | 67.0% | 17.7% |
| aime 25 | 53.7% | 22.0% |
| artificial analysis coding index | 11.40 | — |
| artificial analysis intelligence index | 16.00 | 9.10 |
| artificial analysis math index | 53.70 | 22.00 |
| gpqa | 40.2% | 9.8% |
| hle | 6.1% | 3.3% |
| ifbench | 27.6% | 13.2% |
| lcr | 11.0% | 0.3% |
| livecodebench | 26.6% | 7.0% |
| math 500 | 93.5% | 68.7% |
| mmlu pro | 79.5% | 26.9% |
| scicode | 31.3% | 6.6% |
| tau2 | 21.9% | — |
| terminalbench hard | 1.5% | — |
Benchmark data from Artificial Analysis.