← All comparisons
DeepSeek V4 Flash (Reasoning, High Effort) vs Qwen3.5 9B (Reasoning)
DeepSeek vs Alibaba — side-by-side benchmark comparison
| DeepSeek V4 Flash (Reasoning, High Effort) | Qwen3.5 9B (Reasoning) | |
|---|---|---|
| Intelligence Index | 46.0 | 32.4 |
| Coding Index | 39.8 | 25.3 |
| Math Index | — | — |
| Output speed (tok/s) | 0.0 | 69.4 |
| Blended price ($/1M) | $0.17 | $0.11 |
| Time to first token (s) | 0.00s | 1.37s |
| aime | — | — |
| aime 25 | — | — |
| artificial analysis coding index | 39.80 | 25.30 |
| artificial analysis intelligence index | 46.00 | 32.40 |
| artificial analysis math index | — | — |
| gpqa | 86.7% | 80.6% |
| hle | 27.8% | 13.3% |
| ifbench | 73.5% | 66.7% |
| lcr | 62.7% | 59.0% |
| livecodebench | — | — |
| math 500 | — | — |
| mmlu pro | — | — |
| scicode | 42.0% | 27.5% |
| tau2 | 95.6% | 86.8% |
| terminalbench hard | 38.6% | 24.2% |
Benchmark data from Artificial Analysis.