← All comparisons
Grok 4.20 0309 v2 (Non-reasoning) vs Qwen3 4B 2507 (Reasoning)
xAI vs Alibaba — side-by-side benchmark comparison
| Grok 4.20 0309 v2 (Non-reasoning) | Qwen3 4B 2507 (Reasoning) | |
|---|---|---|
| Intelligence Index | 29.0 | 18.2 |
| Coding Index | 22.0 | 9.5 |
| Math Index | — | 82.7 |
| Output speed (tok/s) | 175.2 | 0.0 |
| Blended price ($/1M) | $3.00 | $0.00 |
| Time to first token (s) | 0.47s | 0.00s |
| aime | — | — |
| aime 25 | — | 82.7% |
| artificial analysis coding index | 22.00 | 9.50 |
| artificial analysis intelligence index | 29.00 | 18.20 |
| artificial analysis math index | — | 82.70 |
| gpqa | 77.6% | 66.7% |
| hle | 24.2% | 5.9% |
| ifbench | 49.3% | 49.8% |
| lcr | 17.3% | 37.7% |
| livecodebench | — | 64.1% |
| math 500 | — | — |
| mmlu pro | — | 74.3% |
| scicode | 32.8% | 25.6% |
| tau2 | 59.9% | 25.4% |
| terminalbench hard | 16.7% | 1.5% |
Benchmark data from Artificial Analysis.