| Benchmark | Claude Sonnet 5 | Kimi K2 (Reasoning) |
|---|---|---|
| AA Agentic Index | 49.7 | 47.9 |
| AA Intelligence | 55.3 | 33.5 |
| AA-LCR | 77 | 70.3 |
| AA-Omniscience | 16.4 | -21.4 |
| Artificial Analysis Coding Index | 71.5 | 34.8 |
| browsecomp | 84.7 | 60.2 |
| critpt | 16.9 | 2.6 |
| gdpval | 54.8 | 24.5 |
| GPQA Diamond | 91.1 | 84.5 |
| HLE | 57.4 | 23.9 |
| OmniScience Accuracy | 40 | 30.9 |
| OmniScience Non-Hallucination | 60.6 | 25.8 |
| scicode | 53.6 | 44.8 |
| simplebench | 60.6 | 39.6 |
| SWE-bench Multilingual | 78.3 | 61.1 |
| SWE-bench Verified | 85.2 | 71.3 |
| Terminal-Bench 2.0 | 80.4 | 35.7 |
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.