← All comparisons
DeepSeek R1 Distill Llama 8B vs Qwen3 30B A3B (Reasoning)
DeepSeek vs Alibaba — side-by-side benchmark comparison
| DeepSeek R1 Distill Llama 8B | Qwen3 30B A3B (Reasoning) | |
|---|---|---|
| Intelligence Index | 12.1 | 15.3 |
| Coding Index | — | 11.0 |
| Math Index | 41.3 | 72.3 |
| Output speed (tok/s) | 0.0 | 64.1 |
| Blended price ($/1M) | $0.00 | $0.18 |
| Time to first token (s) | 0.00s | 1.18s |
| aime | 33.3% | 75.3% |
| aime 25 | 41.3% | 72.3% |
| artificial analysis coding index | — | 11.00 |
| artificial analysis intelligence index | 12.10 | 15.30 |
| artificial analysis math index | 41.30 | 72.30 |
| gpqa | 30.2% | 61.6% |
| hle | 4.2% | 6.6% |
| ifbench | 17.6% | 41.5% |
| lcr | 0.0% | 0.0% |
| livecodebench | 23.3% | 50.6% |
| math 500 | 85.3% | 95.9% |
| mmlu pro | 54.3% | 77.7% |
| scicode | 11.9% | 28.5% |
| tau2 | — | 26.0% |
| terminalbench hard | — | 2.3% |
Benchmark data from Artificial Analysis.