← All comparisons
Ministral 3 8B vs Qwen3.5 0.8B (Non-reasoning)
Mistral vs Alibaba — side-by-side benchmark comparison
| Ministral 3 8B | Qwen3.5 0.8B (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 14.8 | 9.9 |
| Coding Index | 10.0 | 1.0 |
| Math Index | 31.7 | — |
| Output speed (tok/s) | 100.9 | 96.3 |
| Blended price ($/1M) | $0.15 | $0.02 |
| Time to first token (s) | 0.37s | 0.26s |
| aime | — | — |
| aime 25 | 31.7% | — |
| artificial analysis coding index | 10.00 | 100.0% |
| artificial analysis intelligence index | 14.80 | 9.90 |
| artificial analysis math index | 31.70 | — |
| gpqa | 47.1% | 23.6% |
| hle | 4.3% | 4.9% |
| ifbench | 29.1% | 21.6% |
| lcr | 24.0% | 6.7% |
| livecodebench | 30.3% | — |
| math 500 | — | — |
| mmlu pro | 64.2% | — |
| scicode | 20.8% | 2.9% |
| tau2 | 26.6% | 65.2% |
| terminalbench hard | 4.5% | 0.0% |
Benchmark data from Artificial Analysis.