| Benchmark | Gemini 1.5 Pro | Kimi K2 (Reasoning) |
|---|---|---|
| AA Intelligence | 10 | 33.5 |
| aider_polyglot | 16.9 | 59.1 |
| Artificial Analysis Coding Index | 23.6 | 34.8 |
| GPQA Diamond | 59.1 | 84.5 |
| HLE | 4.9 | 23.9 |
| LiveCodeBench | 30.5 | 79.2 |
| MMLU-Pro | 75.8 | 84.6 |
| mrcr | 82.6 | 44.2 |
| scicode | 29.5 | 44.8 |
| simplebench | 27.1 | 39.6 |
| SWE-bench Verified | 34.2 | 71.3 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.