| Benchmark | GPT-4o mini | Kimi K2 (Reasoning) |
|---|---|---|
| AA Agentic Index | 1 | 47.9 |
| AA Intelligence | 7 | 33.5 |
| aider_polyglot | 3.6 | 59.1 |
| Artificial Analysis Coding Index | 11.4 | 34.8 |
| gdpval | 0 | 24.5 |
| GPQA Diamond | 42.6 | 84.5 |
| HLE | 4.2 | 23.9 |
| IFBench | 31 | 68.1 |
| LiveCodeBench | 35.5 | 79.2 |
| scicode | 22.9 | 44.8 |
| simplebench | 10.7 | 39.6 |
| SWE-bench Verified | 8.7 | 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.