| Benchmark | Claude Sonnet 5 | Kimi K3 |
|---|---|---|
| AA Agentic Index | 49.7 | 54.3 |
| AA Intelligence | 55.3 | 60 |
| AA-LCR | 77 | 82.7 |
| AA-Omniscience | 16.4 | 46 |
| arena_elo | 1463 | 1489 |
| arena_text_factuality | 1453 | 1471 |
| Artificial Analysis Coding Index | 71.5 | 76.2 |
| browsecomp | 84.7 | 91.2 |
| charxiv_rq | 77 | 84.8 |
| coding_arena_elo | 1540 | 1674 |
| critpt | 16.9 | 23.4 |
| DeepSWE 1.1 | 54 | 69 |
| gdpval | 54.8 | 59.4 |
| GDPval-AA v2 | 1618 | 1687 |
| GPQA Diamond | 91.1 | 93.5 |
| HLE | 57.4 | 56 |
| livebench_agentic_coding | 59.4 | 62.2 |
| livebench_coding | 80.7 | 81.5 |
| livebench_data_analysis | 71.7 | 78.7 |
| livebench_instruction_following | 63.9 | 71.4 |
| livebench_language | 75 | 85.5 |
| livebench_math | 92.9 | 84.4 |
| livebench_reasoning | 88.7 | 90.7 |
| MMMU-Pro | 77.3 | 83.4 |
| officeqa_pro | 59.4 | 63.3 |
| OmniScience Accuracy | 40 | 47.6 |
| OmniScience Non-Hallucination | 60.6 | 49.1 |
| OSWorld-Verified | 81.2 | 84.8 |
| scicode | 53.6 | 58.7 |
| simplebench | 60.6 | 60.7 |
| simpleqa_verified | 25 | 42.7 |
| TauBench V3 - Banking | 37.3 | 46 |
| Terminal-Bench 2.1 | 80.5 | 88.3 |
| toolathlon | 54.3 | 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.