| Benchmark | Kimi K3 | o3 |
|---|---|---|
| AA Agentic Index | 54.3 | 36.1 |
| AA Intelligence | 60 | 31.1 |
| AA-LCR | 82.7 | 73.3 |
| AA-Omniscience | 46 | -15.3 |
| arc_agi_1 | 65.7 | 60.8 |
| ARC-AGI-2 | 12.4 | 6.5 |
| Artificial Analysis Coding Index | 76.2 | 38.4 |
| browsecomp | 91.2 | 49.7 |
| critpt | 23.4 | 1.1 |
| gdpval | 59.4 | 12.8 |
| GPQA Diamond | 93.5 | 83.3 |
| HLE | 56 | 20.3 |
| MMMU-Pro | 83.4 | 76.4 |
| OmniScience Accuracy | 47.6 | 38.6 |
| OmniScience Non-Hallucination | 49.1 | 12.9 |
| scicode | 58.7 | 41 |
| simplebench | 60.7 | 53.1 |
| simpleqa_verified | 42.7 | 53 |
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.