| Benchmark | GPT-4o mini | Kimi K3 |
|---|---|---|
| AA Agentic Index | 1 | 54.3 |
| AA Intelligence | 7 | 60 |
| ARC-AGI-2 | 0 | 12.4 |
| Artificial Analysis Coding Index | 11.4 | 76.2 |
| gdpval | 0 | 59.4 |
| GPQA Diamond | 42.6 | 93.5 |
| HLE | 4.2 | 56 |
| MMMU-Pro | 41.5 | 83.4 |
| scicode | 22.9 | 58.7 |
| simplebench | 10.7 | 60.7 |
| TauBench V3 - Banking | 2.9 | 46 |
| Terminal-Bench 2.1 | 5.6 | 88.3 |
| Video-MME | 68.9 | 90 |
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.