← All comparisons
DeepSeek V3.2 Exp (Reasoning) vs Qwen3 235B A22B (Reasoning)
DeepSeek vs Alibaba — side-by-side benchmark comparison
| DeepSeek V3.2 Exp (Reasoning) | Qwen3 235B A22B (Reasoning) | |
|---|---|---|
| Intelligence Index | 32.9 | 19.8 |
| Coding Index | 33.3 | 17.4 |
| Math Index | 87.7 | 82.0 |
| Output speed (tok/s) | 0.0 | 58.3 |
| Blended price ($/1M) | $0.31 | $2.63 |
| Time to first token (s) | 0.00s | 1.37s |
| aime | — | 84.0% |
| aime 25 | 87.7% | 82.0% |
| artificial analysis coding index | 33.30 | 17.40 |
| artificial analysis intelligence index | 32.90 | 19.80 |
| artificial analysis math index | 87.70 | 82.00 |
| gpqa | 79.7% | 70.0% |
| hle | 13.8% | 11.7% |
| ifbench | 54.1% | 38.7% |
| lcr | 69.0% | 0.0% |
| livecodebench | 78.9% | 62.2% |
| math 500 | — | 93.0% |
| mmlu pro | 85.0% | 82.8% |
| scicode | 37.7% | 39.9% |
| tau2 | 33.9% | 24.0% |
| terminalbench hard | 31.1% | 6.1% |
Benchmark data from Artificial Analysis.