| Benchmark | Kimi K2.6 | o4-mini |
|---|---|---|
| AA Agentic Index | 58.7 | 36.1 |
| AA Intelligence | 45.1 | 26.1 |
| AA-LCR | 76.7 | 60 |
| AA-Omniscience | 6.4 | -35.7 |
| arena_vision | 1264 | 1201 |
| Artificial Analysis Coding Index | 61.8 | 25.6 |
| browsecomp | 86.3 | 51.5 |
| critpt | 8 | 0.6 |
| frontiermath_tier_4 | 14.6 | 6.3 |
| gdpval | 41.4 | 25.4 |
| GPQA Diamond | 91.1 | 81.4 |
| HLE | 37.5 | 16.5 |
| IFBench | 76 | 68.7 |
| LiveCodeBench | 89.6 | 84.5 |
| MMMU-Pro | 80.1 | 69.2 |
| OmniScience Accuracy | 32.8 | 24.8 |
| OmniScience Non-Hallucination | 60.7 | 19.5 |
| scicode | 53.5 | 46.5 |
| simpleqa_verified | 38.7 | 23.9 |
| strongreject | 99.8 | 96 |
| SWE-bench Verified | 80.2 | 68.1 |
| Terminal-Bench Hard | 43.9 | 15.2 |
| vectara_answer_rate | 99.7 | 99.2 |
| vectara_avg_summary_length | 116.7 | 130.9 |
| vectara_factual_consistency | 89.2 | 81.4 |
| vectara_hallucination_rate ↓ | 10.8 | 18.6 |
| τ²-Bench Telecom (AA run) | 95.9 | 55.6 |
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.