| Benchmark | Kimi K2 (Reasoning) | Qwen3.7 Max |
|---|---|---|
| AA Agentic Index | 47.9 | 30.9 |
| AA Intelligence | 33.5 | 46.7 |
| AA-LCR | 70.3 | 74.7 |
| AA-Omniscience | -21.4 | 13.5 |
| Artificial Analysis Coding Index | 34.8 | 66 |
| critpt | 2.6 | 13.4 |
| gdpval | 24.5 | 38.5 |
| GPQA Diamond | 84.5 | 92.4 |
| HealthBench | 58 | 54.5 |
| HLE | 23.9 | 41.4 |
| HLE (with tools) | 44.9 | 53.5 |
| hmmt_nov_2025 | 89.2 | 95 |
| IFBench | 68.1 | 80.5 |
| imo_answer_bench | 78.6 | 90 |
| LiveCodeBench v6 | 83.1 | 91.6 |
| longbench_v2 | 45.1 | 65.3 |
| mmlu_redux | 94.4 | 95 |
| MMLU-Pro | 84.6 | 89.6 |
| OmniScience Accuracy | 30.9 | 31.1 |
| OmniScience Non-Hallucination | 25.8 | 74.4 |
| scicode | 44.8 | 53.5 |
| simplebench | 39.6 | 70.4 |
| SWE-bench Multilingual | 61.1 | 78.3 |
| SWE-bench Verified | 71.3 | 80.4 |
| Terminal-Bench 2.0 | 35.7 | 69.7 |
| Terminal-Bench Hard | 31.1 | 50.8 |
| τ²-Bench Telecom (AA run) | 93 | 94.7 |
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.