| Benchmark | Claude Sonnet 4.6 | Kimi K3 |
|---|---|---|
| AA Agentic Index | 61.6 | 54.3 |
| AA Intelligence | 48.4 | 60 |
| AA-LCR | 74 | 82.7 |
| AA-Omniscience | 12.2 | 46 |
| apexAgents | 28 | 41.3 |
| arc_agi_1 | 86.5 | 65.7 |
| ARC-AGI-2 | 60.4 | 12.4 |
| arena_elo | 1472 | 1489 |
| arena_text_factuality | 1460 | 1471 |
| Artificial Analysis Coding Index | 63 | 76.2 |
| browsecomp | 76.2 | 91.2 |
| charxiv_rq | 72.4 | 84.8 |
| coding_arena_elo | 1523 | 1674 |
| critpt | 3.1 | 23.4 |
| DeepSWE 1.1 | 30 | 69 |
| Finance Agent v2 | 51 | 54.4 |
| gdpval | 54.8 | 59.4 |
| GDPval-AA v2 | 1381 | 1687 |
| GPQA Diamond | 89.9 | 93.5 |
| HLE | 49 | 56 |
| itbenchSre | 39.8 | 47.7 |
| livebench_agentic_coding | 42.6 | 62.2 |
| livebench_coding | 79.3 | 81.5 |
| livebench_data_analysis | 78 | 78.7 |
| livebench_instruction_following | 63.2 | 71.4 |
| livebench_language | 76.1 | 85.5 |
| livebench_math | 87 | 84.4 |
| livebench_reasoning | 84.8 | 90.7 |
| MCP Atlas | 69.5 | 84.2 |
| MMMU-Pro | 75.6 | 83.4 |
| officeqa_pro | 53.4 | 63.3 |
| OmniScience Accuracy | 40.9 | 47.6 |
| OmniScience Non-Hallucination | 51.6 | 49.1 |
| OSWorld-Verified | 78.5 | 84.8 |
| scicode | 47 | 58.7 |
| simpleqa_verified | 29 | 42.7 |
| TauBench V3 - Banking | 34.4 | 46 |
| Terminal-Bench 2.1 | 71.2 | 88.3 |
| toolathlon | 49.4 | 73.2 |
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.