| Benchmark | GPT-5.5 | Qwen3.6 27B |
|---|---|---|
| AA Agentic Index | 47.4 | 27.5 |
| AA Intelligence | 56.3 | 37.7 |
| AA-LCR | 79 | 73.3 |
| AA-Omniscience | 20.5 | -20 |
| Agents' Last Exam | 26.6 | 27.3 |
| aime_2026 | 100 | 94.1 |
| Artificial Analysis Coding Index | 74.9 | 53.7 |
| critpt | 27.1 | 1.1 |
| DeepSWE 1.1 | 67 | 13.3 |
| gdpval | 49.5 | 32 |
| GPQA Diamond | 93.6 | 87.8 |
| HLE | 52.2 | 24 |
| hmmt_feb_2026 | 98.5 | 84.3 |
| hmmt_nov_2025 | 96.5 | 90.7 |
| IFBench | 75.9 | 69.1 |
| JobBench | 38.3 | 21.8 |
| livebench_agentic_coding | 54 | 39.3 |
| livebench_coding | 82.2 | 71.8 |
| livebench_data_analysis | 81.6 | 70.4 |
| livebench_instruction_following | 70.7 | 53.2 |
| livebench_language | 87.4 | 63.3 |
| livebench_math | 95.9 | 79.9 |
| livebench_reasoning | 89.7 | 70.3 |
| MMMU-Pro | 83.2 | 75.8 |
| nl2repo | 50.7 | 36.2 |
| OmniScience Accuracy | 58 | 19.6 |
| OmniScience Non-Hallucination | 12 | 50.7 |
| OSWorld-Verified | 78.7 | 63.9 |
| scicode | 56.1 | 39.8 |
| SWE-bench Pro | 58.6 | 53.5 |
| SWE-bench Verified | 80.6 | 77.2 |
| TauBench V3 - Banking | 39 | 16.7 |
| Terminal-Bench 2.0 | 82.7 | 59.3 |
| Terminal-Bench 2.1 | 84.3 | 60.7 |
| Terminal-Bench Hard | 60.6 | 34.8 |
| τ²-Bench Telecom (AA run) | 93.9 | 94.2 |
| τ³-Bench | 31.3 | 15.3 |
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.