| Benchmark | Kimi K2.5 | o1 |
|---|---|---|
| AA Agentic Index | 52.8 | 31.1 |
| AA Intelligence | 36 | 24 |
| AA-LCR | 73 | 63.3 |
| AA-Omniscience | -7.3 | -10.6 |
| AIME 2025 | 96.1 | 81.7 |
| arena_vision | 1247 | 1193 |
| Artificial Analysis Coding Index | 46.8 | 39.7 |
| critpt | 3.1 | 0.3 |
| Fortress | 41.1 | 19.4 |
| gdpval | 38.3 | 11.5 |
| GPQA Diamond | 87.9 | 75.7 |
| HLE | 50.2 | 7.7 |
| IFBench | 70.2 | 70.3 |
| livebench | 69.1 | 52.3 |
| MathVista | 90.1 | 71.8 |
| OmniScience Accuracy | 35.2 | 34.7 |
| OmniScience Non-Hallucination | 50 | 30.7 |
| scicode | 49 | 35.8 |
| SEAL VISTA | 41.9 | 45.3 |
| simplebench | 46.8 | 41.7 |
| strongreject | 99.5 | 97 |
| SWE-bench Verified | 76.8 | 48.9 |
| Terminal-Bench Hard | 34.8 | 12.9 |
| τ²-Bench Telecom (AA run) | 95.9 | 62.6 |
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.