| Benchmark | Claude 4 Opus | Qwen3.6 27B |
|---|---|---|
| AA Intelligence | 31.7 | 37.7 |
| AA-LCR | 40 | 73.3 |
| Artificial Analysis Coding Index | 34 | 53.7 |
| GPQA Diamond | 79.6 | 87.8 |
| HLE | 12.3 | 24 |
| HMMT 2025 | 15.9 | 90.7 |
| hmmt_feb_2025 | 60 | 93.8 |
| IFBench | 53.7 | 69.1 |
| LiveCodeBench v6 | 47.4 | 83.9 |
| mmlu_redux | 94.2 | 93.5 |
| MMLU-Pro | 86.6 | 86.2 |
| MMMU | 73.7 | 82.9 |
| scicode | 40.9 | 39.8 |
| simpleqa | 22.8 | 95.7 |
| supergpqa | 56.5 | 66 |
| SWE-bench Verified | 79.4 | 77.2 |
| Terminal-Bench Hard | 31.1 | 34.8 |
| τ²-Bench Telecom (AA run) | 73.4 | 94.2 |
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.