| Benchmark | Kimi K2.6 | Opus 5 |
|---|---|---|
| AA Agentic Index | 58.7 | 59.2 |
| AA Intelligence | 45.1 | 63.1 |
| AA-LCR | 76.7 | 78.7 |
| AA-Omniscience | 6.4 | 37.1 |
| arena_elo | 1461 | 1487 |
| arena_text_factuality | 1455 | 1481 |
| Artificial Analysis Coding Index | 61.8 | 78 |
| browsecomp | 86.3 | 90.8 |
| coding_arena_elo | 1509 | 1662 |
| critpt | 8 | 29.1 |
| Finance Agent v2 | 44.9 | 58.6 |
| gdpval | 41.4 | 67.2 |
| GDPval-AA (Elo) | 1482 | 1827 |
| GDPval-AA v2 | 1190 | 1861 |
| GPQA Diamond | 91.1 | 93.7 |
| HiL-Bench | 18.7 | 57 |
| HLE | 37.5 | 64.7 |
| HLE (with tools) | 54 | 64.7 |
| livebench_agentic_coding | 46.9 | 65.2 |
| livebench_coding | 78.6 | 81.5 |
| livebench_data_analysis | 65.1 | 74.5 |
| livebench_instruction_following | 64.4 | 63.8 |
| livebench_language | 75.1 | 88.7 |
| livebench_math | 84.3 | 95.7 |
| livebench_reasoning | 79.4 | 91.2 |
| MCP Atlas | 68.1 | 85.8 |
| MMMU-Pro | 80.1 | 84.7 |
| OmniScience Accuracy | 32.8 | 60.9 |
| OmniScience Non-Hallucination | 60.7 | 40.5 |
| OSWorld-Verified | 73.1 | 4 |
| scicode | 53.5 | 55.7 |
| simpleqa_verified | 38.7 | 56.7 |
| SWE-bench Multilingual | 76.7 | 89.5 |
| SWE-bench Pro | 58.6 | 79.2 |
| SWE-bench Verified | 80.2 | 96 |
| TauBench V3 - Banking | 23.3 | 44.7 |
| Terminal-Bench 2.1 | 65.9 | 89.1 |
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.