← All comparisons
Qwen3 4B (Non-reasoning) vs Qwen3 1.7B (Non-reasoning)
Alibaba vs Alibaba — side-by-side benchmark comparison
| Qwen3 4B (Non-reasoning) | Qwen3 1.7B (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 12.5 | 6.8 |
| Coding Index | — | 2.3 |
| Math Index | — | 7.3 |
| Output speed (tok/s) | 103.5 | 137.6 |
| Blended price ($/1M) | $0.19 | $0.19 |
| Time to first token (s) | 1.02s | 0.93s |
| aime | 21.3% | 9.7% |
| aime 25 | — | 7.3% |
| artificial analysis coding index | — | 2.30 |
| artificial analysis intelligence index | 12.50 | 6.80 |
| artificial analysis math index | — | 7.30 |
| gpqa | 39.8% | 28.3% |
| hle | 3.7% | 5.2% |
| ifbench | — | 21.1% |
| lcr | — | 0.0% |
| livecodebench | 23.3% | 12.6% |
| math 500 | 84.3% | 71.7% |
| mmlu pro | 58.6% | 41.1% |
| scicode | 16.7% | 6.9% |
| tau2 | — | 21.6% |
| terminalbench hard | — | 0.0% |
Benchmark data from Artificial Analysis.