| Benchmark | Kimi K2.5 | Qwen3.7 Max |
|---|---|---|
| AA Agentic Index | 52.8 | 30.9 |
| AA Intelligence | 36 | 46.7 |
| AA-LCR | 73 | 74.7 |
| AA-Omniscience | -7.3 | 13.5 |
| aime_2026 | 95.8 | 97 |
| arena_elo | 1450 | 1474 |
| arena_text_factuality | 1445 | 1479 |
| Artificial Analysis Coding Index | 46.8 | 66 |
| coding_arena_elo | 1436 | 1517 |
| critpt | 3.1 | 13.4 |
| gdpval | 38.3 | 38.5 |
| GPQA Diamond | 87.9 | 92.4 |
| HLE | 50.2 | 41.4 |
| HLE (with tools) | 51.8 | 53.5 |
| hmmt_feb_2026 | 87.1 | 97.1 |
| hmmt_nov_2025 | 91.1 | 95 |
| IFBench | 70.2 | 80.5 |
| imo_answer_bench | 81.8 | 90 |
| livebench | 69.1 | 74.3 |
| LiveCodeBench v6 | 85 | 91.6 |
| longbench_v2 | 61 | 65.3 |
| MCP Atlas | 64 | 76.4 |
| mcpmark | 29.5 | 60.8 |
| MMLU-Pro | 87.1 | 89.6 |
| nl2repo | 32 | 47.2 |
| OmniScience Accuracy | 35.2 | 31.1 |
| OmniScience Non-Hallucination | 50 | 74.4 |
| PaperBench | 63.5 | 64.8 |
| scicode | 49 | 53.5 |
| simplebench | 46.8 | 70.4 |
| simpleqa_verified | 36.9 | 58.5 |
| SWE-bench Multilingual | 73 | 78.3 |
| SWE-bench Pro | 53.8 | 60.6 |
| SWE-bench Verified | 76.8 | 80.4 |
| TauBench V3 - Banking | 14.2 | 11.8 |
| Terminal-Bench 2.0 | 50.8 | 69.7 |
| Terminal-Bench 2.1 | 45.7 | 75 |
| Terminal-Bench Hard | 34.8 | 50.8 |
| WideSearch | 79 | 75.2 |
| τ²-Bench Telecom (AA run) | 95.9 | 94.7 |
| τ³-Bench | 66 | 10.9 |
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.