← All comparisons
Qwen3.5 122B A10B (Non-reasoning) vs Qwen3 235B A22B 2507 Instruct
Alibaba vs Alibaba — side-by-side benchmark comparison
| Qwen3.5 122B A10B (Non-reasoning) | Qwen3 235B A22B 2507 Instruct | |
|---|---|---|
| Intelligence Index | 35.9 | 25.0 |
| Coding Index | 31.6 | 22.1 |
| Math Index | — | 71.7 |
| Output speed (tok/s) | 165.9 | 57.0 |
| Blended price ($/1M) | $1.10 | $0.36 |
| Time to first token (s) | 1.06s | 1.34s |
| aime | — | 71.7% |
| aime 25 | — | 71.7% |
| artificial analysis coding index | 31.60 | 22.10 |
| artificial analysis intelligence index | 35.90 | 25.00 |
| artificial analysis math index | — | 71.70 |
| gpqa | 82.7% | 75.3% |
| hle | 14.8% | 10.6% |
| ifbench | 50.8% | 46.1% |
| lcr | 56.0% | 31.2% |
| livecodebench | — | 52.4% |
| math 500 | — | 98.0% |
| mmlu pro | — | 82.8% |
| scicode | 35.6% | 36.0% |
| tau2 | 84.5% | 33.3% |
| terminalbench hard | 29.5% | 15.2% |
Benchmark data from Artificial Analysis.