| Benchmark | Opus 5 | Qwen3.6 Plus |
|---|---|---|
| AA Agentic Index | 59.2 | 29 |
| AA Intelligence | 63.1 | 40.5 |
| AA-LCR | 78.7 | 72.3 |
| AA-Omniscience | 37.1 | 2.6 |
| arena_elo | 1487 | 1444 |
| arena_text_factuality | 1481 | 1438 |
| Artificial Analysis Coding Index | 78 | 54.5 |
| coding_arena_elo | 1662 | 1460 |
| critpt | 29.1 | 2.9 |
| Finance Agent v2 | 58.6 | 40.9 |
| gdpval | 67.2 | 32 |
| GPQA Diamond | 93.7 | 90.4 |
| HLE | 64.7 | 28.8 |
| HLE (with tools) | 64.7 | 50.6 |
| livebench_agentic_coding | 65.2 | 41.4 |
| livebench_coding | 81.5 | 78.2 |
| livebench_data_analysis | 74.5 | 69.9 |
| livebench_instruction_following | 63.8 | 58.3 |
| livebench_language | 88.7 | 75 |
| livebench_math | 95.7 | 83.7 |
| livebench_reasoning | 91.2 | 75.8 |
| MCP Atlas | 85.8 | 74.1 |
| MMMU-Pro | 84.7 | 78.8 |
| OmniScience Accuracy | 60.9 | 26.4 |
| OmniScience Non-Hallucination | 40.5 | 68 |
| OSWorld-Verified | 4 | 62.5 |
| scicode | 55.7 | 40.7 |
| simpleqa_verified | 56.7 | 49.1 |
| SWE-bench Multilingual | 89.5 | 73.8 |
| SWE-bench Pro | 79.2 | 56.6 |
| SWE-bench Verified | 96 | 78.8 |
| TauBench V3 - Banking | 44.7 | 20.8 |
| Terminal-Bench 2.1 | 89.1 | 61.4 |
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.