← All comparisons
MiniMax-M2.5 vs Qwen3 4B (Non-reasoning)
MiniMax vs Alibaba — side-by-side benchmark comparison
| MiniMax-M2.5 | Qwen3 4B (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 41.9 | 12.5 |
| Coding Index | 37.4 | — |
| Math Index | — | — |
| Output speed (tok/s) | 190.8 | 103.5 |
| Blended price ($/1M) | $0.53 | $0.19 |
| Time to first token (s) | 2.38s | 1.02s |
| aime | — | 21.3% |
| aime 25 | — | — |
| artificial analysis coding index | 37.40 | — |
| artificial analysis intelligence index | 41.90 | 12.50 |
| artificial analysis math index | — | — |
| gpqa | 84.8% | 39.8% |
| hle | 19.1% | 3.7% |
| ifbench | 71.6% | — |
| lcr | 66.0% | — |
| livecodebench | — | 23.3% |
| math 500 | — | 84.3% |
| mmlu pro | — | 58.6% |
| scicode | 42.6% | 16.7% |
| tau2 | 95.3% | — |
| terminalbench hard | 34.8% | — |
Benchmark data from Artificial Analysis.