← All comparisons
GPT-5.4 (Non-reasoning) vs Qwen3 VL 30B A3B Instruct
OpenAI vs Alibaba — side-by-side benchmark comparison
| GPT-5.4 (Non-reasoning) | Qwen3 VL 30B A3B Instruct | |
|---|---|---|
| Intelligence Index | 35.4 | 16.0 |
| Coding Index | 41.0 | 14.3 |
| Math Index | — | 72.3 |
| Output speed (tok/s) | 70.9 | 123.5 |
| Blended price ($/1M) | $5.63 | $0.30 |
| Time to first token (s) | 0.60s | 1.07s |
| aime | — | — |
| aime 25 | — | 72.3% |
| artificial analysis coding index | 41.00 | 14.30 |
| artificial analysis intelligence index | 35.40 | 16.00 |
| artificial analysis math index | — | 72.30 |
| gpqa | 74.8% | 69.5% |
| hle | 10.6% | 6.4% |
| ifbench | 48.4% | 33.1% |
| lcr | 47.3% | 23.7% |
| livecodebench | — | 47.6% |
| math 500 | — | — |
| mmlu pro | — | 76.4% |
| scicode | 47.1% | 30.8% |
| tau2 | 35.1% | 19.0% |
| terminalbench hard | 37.9% | 6.1% |
Benchmark data from Artificial Analysis.