← All comparisons
Gemma 4 26B A4B (Non-reasoning) vs Qwen3 VL 4B (Reasoning)
Google vs Alibaba — side-by-side benchmark comparison
| Gemma 4 26B A4B (Non-reasoning) | Qwen3 VL 4B (Reasoning) | |
|---|---|---|
| Intelligence Index | 27.1 | 13.7 |
| Coding Index | 29.1 | 6.7 |
| Math Index | — | 25.7 |
| Output speed (tok/s) | 71.1 | 0.0 |
| Blended price ($/1M) | $0.20 | $0.00 |
| Time to first token (s) | 0.80s | 0.00s |
| aime | — | — |
| aime 25 | — | 25.7% |
| artificial analysis coding index | 29.10 | 6.70 |
| artificial analysis intelligence index | 27.10 | 13.70 |
| artificial analysis math index | — | 25.70 |
| gpqa | 71.4% | 49.4% |
| hle | 10.7% | 4.4% |
| ifbench | 45.4% | 36.6% |
| lcr | 39.7% | 21.3% |
| livecodebench | — | 32.0% |
| math 500 | — | — |
| mmlu pro | — | 70.0% |
| scicode | 37.3% | 17.1% |
| tau2 | 40.4% | 15.5% |
| terminalbench hard | 25.0% | 1.5% |
Benchmark data from Artificial Analysis.