← All comparisons
Qwen3 235B A22B 2507 (Reasoning) vs Qwen3 30B A3B 2507 Instruct
Alibaba vs Alibaba — side-by-side benchmark comparison
| Qwen3 235B A22B 2507 (Reasoning) | Qwen3 30B A3B 2507 Instruct | |
|---|---|---|
| Intelligence Index | 29.5 | 15.0 |
| Coding Index | 23.2 | 14.2 |
| Math Index | 91.0 | 66.3 |
| Output speed (tok/s) | 62.5 | 102.1 |
| Blended price ($/1M) | $0.84 | $0.21 |
| Time to first token (s) | 1.21s | 0.98s |
| aime | 94.0% | 72.7% |
| aime 25 | 91.0% | 66.3% |
| artificial analysis coding index | 23.20 | 14.20 |
| artificial analysis intelligence index | 29.50 | 15.00 |
| artificial analysis math index | 91.00 | 66.30 |
| gpqa | 79.0% | 65.9% |
| hle | 15.0% | 6.8% |
| ifbench | 51.2% | 33.1% |
| lcr | 67.0% | 22.7% |
| livecodebench | 78.8% | 51.5% |
| math 500 | 98.4% | 97.5% |
| mmlu pro | 84.3% | 77.7% |
| scicode | 42.4% | 30.4% |
| tau2 | 53.2% | 10.2% |
| terminalbench hard | 13.6% | 6.1% |
Benchmark data from Artificial Analysis.