← All comparisons
Ministral 3 14B vs Gemini 2.5 Flash-Lite Preview (Sep '25) (Non-reasoning)
Mistral vs Google — side-by-side benchmark comparison
| Ministral 3 14B | Gemini 2.5 Flash-Lite Preview (Sep '25) (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 16.0 | 19.4 |
| Coding Index | 10.9 | 14.5 |
| Math Index | 30.0 | 46.7 |
| Output speed (tok/s) | 90.6 | 0.0 |
| Blended price ($/1M) | $0.20 | $0.17 |
| Time to first token (s) | 0.37s | 0.00s |
| aime | — | — |
| aime 25 | 30.0% | 46.7% |
| artificial analysis coding index | 10.90 | 14.50 |
| artificial analysis intelligence index | 16.00 | 19.40 |
| artificial analysis math index | 30.00 | 46.70 |
| gpqa | 57.2% | 65.1% |
| hle | 4.6% | 4.6% |
| ifbench | 32.0% | 41.8% |
| lcr | 22.0% | 48.0% |
| livecodebench | 35.1% | 64.1% |
| math 500 | — | — |
| mmlu pro | 69.3% | 79.6% |
| scicode | 23.6% | 28.5% |
| tau2 | 27.2% | 30.4% |
| terminalbench hard | 4.5% | 7.6% |
Benchmark data from Artificial Analysis.