| Benchmark | Claude 4 Opus | Kimi K3 |
|---|---|---|
| AA Intelligence | 31.7 | 60 |
| AA-LCR | 40 | 82.7 |
| arc_agi_1 | 27 | 65.7 |
| ARC-AGI-2 | 8.6 | 12.4 |
| Artificial Analysis Coding Index | 34 | 76.2 |
| GPQA Diamond | 79.6 | 93.5 |
| HLE | 12.3 | 56 |
| scicode | 40.9 | 58.7 |
| simplebench | 58.8 | 60.7 |
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.