← All comparisons
Qwen3.5 4B (Non-reasoning) vs Grok 3 mini Reasoning (high)
Alibaba vs xAI — side-by-side benchmark comparison
| Qwen3.5 4B (Non-reasoning) | Grok 3 mini Reasoning (high) | |
|---|---|---|
| Intelligence Index | 22.6 | 32.1 |
| Coding Index | 13.7 | 25.2 |
| Math Index | — | 84.7 |
| Output speed (tok/s) | 210.0 | 56.8 |
| Blended price ($/1M) | $0.06 | $0.35 |
| Time to first token (s) | 0.23s | 0.42s |
| aime | — | 93.3% |
| aime 25 | — | 84.7% |
| artificial analysis coding index | 13.70 | 25.20 |
| artificial analysis intelligence index | 22.60 | 32.10 |
| artificial analysis math index | — | 84.70 |
| gpqa | 71.2% | 79.1% |
| hle | 7.5% | 11.1% |
| ifbench | 33.3% | 45.9% |
| lcr | 28.3% | 50.3% |
| livecodebench | — | 69.6% |
| math 500 | — | 99.2% |
| mmlu pro | — | 82.8% |
| scicode | 18.3% | 40.6% |
| tau2 | 87.7% | 90.4% |
| terminalbench hard | 11.4% | 17.4% |
Benchmark data from Artificial Analysis.