← All comparisons
Qwen3 Next 80B A3B (Reasoning) vs Qwen3.5 397B A17B (Non-reasoning)
Alibaba vs Alibaba — side-by-side benchmark comparison
| Qwen3 Next 80B A3B (Reasoning) | Qwen3.5 397B A17B (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 26.7 | 40.1 |
| Coding Index | 19.5 | 37.4 |
| Math Index | 84.3 | — |
| Output speed (tok/s) | 149.4 | 53.5 |
| Blended price ($/1M) | $1.88 | $1.35 |
| Time to first token (s) | 1.23s | 1.85s |
| aime | — | — |
| aime 25 | 84.3% | — |
| artificial analysis coding index | 19.50 | 37.40 |
| artificial analysis intelligence index | 26.70 | 40.10 |
| artificial analysis math index | 84.30 | — |
| gpqa | 75.9% | 86.1% |
| hle | 11.7% | 18.8% |
| ifbench | 60.7% | 51.6% |
| lcr | 60.3% | 58.0% |
| livecodebench | 78.4% | — |
| math 500 | — | — |
| mmlu pro | 82.4% | — |
| scicode | 38.8% | 41.1% |
| tau2 | 41.5% | 83.9% |
| terminalbench hard | 9.8% | 35.6% |
Benchmark data from Artificial Analysis.