| Benchmark | Gemini 2.5 Flash | Kimi K2.6 |
|---|---|---|
| AA Agentic Index | 18.8 | 58.7 |
| AA Intelligence | 17.7 | 45.1 |
| AA-LCR | 65.7 | 76.7 |
| AA-Omniscience | -29.8 | 6.4 |
| arena_text_factuality | 1437 | 1455 |
| arena_vision | 1214 | 1264 |
| Artificial Analysis Coding Index | 22.2 | 61.8 |
| critpt | 1.4 | 8 |
| frontiermath_tier_4 | 4.2 | 14.6 |
| gdpval | 11.9 | 41.4 |
| Global-MMLU-Lite | 88.4 | 88.4 |
| GPQA Diamond | 82.8 | 91.1 |
| HLE | 12.1 | 37.5 |
| IFBench | 50.3 | 76 |
| LiveCodeBench | 76.2 | 89.6 |
| MMMU-Pro | 69.1 | 80.1 |
| OmniScience Accuracy | 26.1 | 32.8 |
| OmniScience Non-Hallucination | 24.7 | 60.7 |
| scicode | 35.9 | 53.5 |
| SWE-bench Verified | 60.4 | 80.2 |
| Terminal-Bench Hard | 13.6 | 43.9 |
| vectara_answer_rate | 99 | 99.7 |
| vectara_avg_summary_length | 101.5 | 116.7 |
| vectara_factual_consistency | 92.2 | 89.2 |
| vectara_hallucination_rate ↓ | 7.8 | 10.8 |
| τ²-Bench Telecom (AA run) | 31.6 | 95.9 |
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.