| Benchmark | GPT-5.1 | Qwen3.6 Plus |
|---|---|---|
| AA Agentic Index | 32.2 | 29 |
| AA Intelligence | 37.5 | 40.5 |
| AA-LCR | 76.7 | 72.3 |
| AA-Omniscience | 5.4 | 2.6 |
| arena_elo | 1439 | 1444 |
| arena_text_factuality | 1445 | 1438 |
| Artificial Analysis Coding Index | 49.4 | 54.5 |
| coding_arena_elo | 1421 | 1460 |
| critpt | 4.9 | 2.9 |
| frontiermath_tier_4 | 12.5 | 8.3 |
| gdpval | 25 | 32 |
| GPQA Diamond | 88.1 | 90.4 |
| HLE | 28.5 | 28.8 |
| HLE (with tools) | 42.7 | 50.6 |
| hmmt_nov_2025 | 91.7 | 94.6 |
| IFBench | 72.9 | 75.2 |
| LiveCodeBench v6 | 87 | 87.1 |
| MCP Atlas | 50.1 | 74.1 |
| MMLU-Pro | 87 | 88.5 |
| mmmlu | 91 | 89.5 |
| MMMU | 85.4 | 86 |
| MMMU-Pro | 75.5 | 78.8 |
| OmniScience Accuracy | 37.7 | 26.4 |
| OmniScience Non-Hallucination | 48.1 | 68 |
| scicode | 43.3 | 40.7 |
| simpleqa_verified | 48.9 | 49.1 |
| SWE-bench Verified | 76.3 | 78.8 |
| TauBench V3 - Banking | 15.9 | 20.8 |
| Terminal-Bench 2.0 | 47.6 | 61.6 |
| Terminal-Bench 2.1 | 52.4 | 61.4 |
| Terminal-Bench Hard | 45.5 | 43.9 |
| τ²-Bench Telecom (AA run) | 81.9 | 97.7 |
| τ³-Bench | 14 | 70.7 |
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.