| Benchmark | Gemini 2.5 Pro | Kimi K2.6 |
|---|---|---|
| AA Agentic Index | 7.2 | 58.7 |
| AA Intelligence | 27 | 45.1 |
| AA-LCR | 66 | 76.7 |
| AA-Omniscience | -14.3 | 6.4 |
| arena_elo | 1446 | 1461 |
| arena_text_factuality | 1462 | 1455 |
| arena_vision | 1246 | 1264 |
| Artificial Analysis Coding Index | 46.7 | 61.8 |
| browsecomp | 9.9 | 86.3 |
| coding_arena_elo | 1224 | 1509 |
| critpt | 2.6 | 8 |
| frontiermath_tier_4 | 4.2 | 14.6 |
| gdpval | 8.5 | 41.4 |
| Global-MMLU-Lite | 88.6 | 88.4 |
| GPQA Diamond | 84.4 | 91.1 |
| HLE | 22.5 | 37.5 |
| HLE (with tools) | 28.4 | 54 |
| IFBench | 49 | 76 |
| LiveCodeBench | 82.7 | 89.6 |
| mathvision | 73.3 | 93.2 |
| MMLU-Pro | 86 | 87.1 |
| MMMU-Pro | 74.9 | 80.1 |
| OmniScience Accuracy | 39 | 32.8 |
| OmniScience Non-Hallucination | 12.6 | 60.7 |
| scicode | 43 | 53.5 |
| simpleqa_verified | 56 | 38.7 |
| SWE-bench Verified | 67.2 | 80.2 |
| TauBench V3 - Banking | 9.7 | 23.3 |
| Terminal-Bench 2.1 | 28.5 | 65.9 |
| Terminal-Bench Hard | 26.5 | 43.9 |
| vectara_answer_rate | 99.1 | 99.7 |
| vectara_avg_summary_length | 106.4 | 116.7 |
| vectara_factual_consistency | 93 | 89.2 |
| vectara_hallucination_rate ↓ | 7 | 10.8 |
| τ²-Bench Telecom (AA run) | 54.1 | 95.9 |
| τ³-Bench | 9.3 | 20.6 |
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.