| Benchmark | Claude 3.7 Sonnet | Kimi K2.6 |
|---|---|---|
| AA Agentic Index | 37 | 58.7 |
| AA Intelligence | 27.6 | 45.1 |
| AA-LCR | 62.3 | 76.7 |
| AA-Omniscience | -0.7 | 6.4 |
| arena_vision | 1195 | 1264 |
| Artificial Analysis Coding Index | 36.4 | 61.8 |
| critpt | 0.9 | 8 |
| gdpval | 27.4 | 41.4 |
| GPQA Diamond | 84.8 | 91.1 |
| HLE | 9.7 | 37.5 |
| IFBench | 48.3 | 76 |
| MMMU-Pro | 60.1 | 80.1 |
| OmniScience Accuracy | 28.2 | 32.8 |
| OmniScience Non-Hallucination | 60 | 60.7 |
| scicode | 40.3 | 53.5 |
| SWE-bench Verified | 70.3 | 80.2 |
| Terminal-Bench Hard | 21.2 | 43.9 |
| τ²-Bench Telecom (AA run) | 54.7 | 95.9 |
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.