| Benchmark | Claude 3.7 Sonnet | Kimi K2 (Reasoning) |
|---|---|---|
| AA Agentic Index | 37 | 47.9 |
| AA Intelligence | 27.6 | 33.5 |
| AA-LCR | 62.3 | 70.3 |
| AA-Omniscience | -0.7 | -21.4 |
| aider_polyglot | 64.9 | 59.1 |
| AIME 2025 | 54.8 | 94.5 |
| Artificial Analysis Coding Index | 36.4 | 34.8 |
| critpt | 0.9 | 2.6 |
| gdpval | 27.4 | 24.5 |
| GPQA Diamond | 84.8 | 84.5 |
| HLE | 9.7 | 23.9 |
| hmmt_feb_2025 | 31.7 | 93.3 |
| IFBench | 48.3 | 68.1 |
| multichallenge | 51.6 | 66.4 |
| OmniScience Accuracy | 28.2 | 30.9 |
| OmniScience Non-Hallucination | 60 | 25.8 |
| scicode | 40.3 | 44.8 |
| simplebench | 46.4 | 39.6 |
| swe_bench_bash | 52.8 | 63.4 |
| SWE-bench Verified | 70.3 | 71.3 |
| Terminal-Bench | 35.2 | 47.1 |
| Terminal-Bench Hard | 21.2 | 31.1 |
| τ²-Bench Telecom (AA run) | 54.7 | 93 |
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.