← All comparisons
Qwen3.5 35B A3B (Non-reasoning) vs DeepSeek R1 0528 Qwen3 8B
Alibaba vs DeepSeek — side-by-side benchmark comparison
| Qwen3.5 35B A3B (Non-reasoning) | DeepSeek R1 0528 Qwen3 8B | |
|---|---|---|
| Intelligence Index | 30.7 | 16.4 |
| Coding Index | 16.8 | 7.8 |
| Math Index | — | 63.7 |
| Output speed (tok/s) | 147.9 | 0.0 |
| Blended price ($/1M) | $0.69 | $0.00 |
| Time to first token (s) | 1.18s | 0.00s |
| aime | — | 65.0% |
| aime 25 | — | 63.7% |
| artificial analysis coding index | 16.80 | 7.80 |
| artificial analysis intelligence index | 30.70 | 16.40 |
| artificial analysis math index | — | 63.70 |
| gpqa | 81.9% | 61.2% |
| hle | 12.8% | 5.6% |
| ifbench | 44.5% | 19.9% |
| lcr | 55.3% | 13.0% |
| livecodebench | — | 51.3% |
| math 500 | — | 93.2% |
| mmlu pro | — | 73.9% |
| scicode | 29.3% | 20.4% |
| tau2 | 86.3% | 0.0% |
| terminalbench hard | 10.6% | 1.5% |
Benchmark data from Artificial Analysis.