← All comparisons
Claude Sonnet 4.6 (Non-reasoning, High Effort) vs Qwen3 VL 235B A22B (Reasoning)
Anthropic vs Alibaba — side-by-side benchmark comparison
| Claude Sonnet 4.6 (Non-reasoning, High Effort) | Qwen3 VL 235B A22B (Reasoning) | |
|---|---|---|
| Intelligence Index | 44.4 | 27.6 |
| Coding Index | 46.4 | 20.9 |
| Math Index | — | 88.3 |
| Output speed (tok/s) | 55.2 | 35.6 |
| Blended price ($/1M) | $6.56 | $2.17 |
| Time to first token (s) | 1.07s | 5.14s |
| aime | — | — |
| aime 25 | — | 88.3% |
| artificial analysis coding index | 46.40 | 20.90 |
| artificial analysis intelligence index | 44.40 | 27.60 |
| artificial analysis math index | — | 88.30 |
| gpqa | 79.9% | 77.2% |
| hle | 13.2% | 10.1% |
| ifbench | 41.2% | 56.5% |
| lcr | 57.7% | 58.7% |
| livecodebench | — | 64.6% |
| math 500 | — | — |
| mmlu pro | — | 83.6% |
| scicode | 46.9% | 39.9% |
| tau2 | 79.5% | 54.1% |
| terminalbench hard | 46.2% | 11.4% |
Benchmark data from Artificial Analysis.