← All comparisons
DeepSeek V4 Flash (Non-reasoning) vs Qwen3 Max
DeepSeek vs Alibaba — side-by-side benchmark comparison
| DeepSeek V4 Flash (Non-reasoning) | Qwen3 Max | |
|---|---|---|
| Intelligence Index | 36.5 | 31.4 |
| Coding Index | 35.2 | 26.4 |
| Math Index | — | 80.7 |
| Output speed (tok/s) | 127.9 | 45.2 |
| Blended price ($/1M) | $0.17 | $3.05 |
| Time to first token (s) | 0.90s | 2.02s |
| aime | — | — |
| aime 25 | — | 80.7% |
| artificial analysis coding index | 35.20 | 26.40 |
| artificial analysis intelligence index | 36.50 | 31.40 |
| artificial analysis math index | — | 80.70 |
| gpqa | 71.6% | 76.4% |
| hle | 7.0% | 11.1% |
| ifbench | 47.2% | 44.1% |
| lcr | 33.3% | 46.7% |
| livecodebench | — | 76.7% |
| math 500 | — | — |
| mmlu pro | — | 84.1% |
| scicode | 37.3% | 38.3% |
| tau2 | 94.4% | 74.3% |
| terminalbench hard | 34.1% | 20.5% |
Benchmark data from Artificial Analysis.