| Benchmark | GPT-5.5 | Qwen3.6 35B A3B |
|---|---|---|
| AA Agentic Index | 47.4 | 21.6 |
| AA Intelligence | 56.3 | 32.1 |
| AA-LCR | 79 | 66.7 |
| AA-Omniscience | 20.5 | -22.2 |
| aime_2026 | 100 | 92.7 |
| Artificial Analysis Coding Index | 74.9 | 41.9 |
| browsecomp | 84.4 | 48.7 |
| critpt | 27.1 | 0.3 |
| gdpval | 49.5 | 27.8 |
| GDPval-AA v2 | 1509 | 1015 |
| GPQA Diamond | 93.6 | 86 |
| HLE | 52.2 | 22.2 |
| hmmt_feb_2026 | 98.5 | 83.6 |
| hmmt_nov_2025 | 96.5 | 89.1 |
| IFBench | 75.9 | 64.4 |
| MCP Atlas | 82.8 | 62.8 |
| MMMU-Pro | 83.2 | 75.3 |
| nl2repo | 50.7 | 29.4 |
| OmniScience Accuracy | 58 | 18.8 |
| OmniScience Non-Hallucination | 12 | 49.5 |
| scicode | 56.1 | 35.8 |
| SWE-bench Pro | 58.6 | 49.5 |
| SWE-bench Verified | 80.6 | 73.4 |
| TauBench V3 - Banking | 39 | 9.3 |
| Terminal-Bench 2.0 | 82.7 | 51.5 |
| Terminal-Bench 2.1 | 84.3 | 44.9 |
| Terminal-Bench Hard | 60.6 | 34.8 |
| tool_decathlon | 55.6 | 26.9 |
| toolathlon | 55.6 | 26.9 |
| τ²-Bench Telecom (AA run) | 93.9 | 95.3 |
| τ³-Bench | 31.3 | 67.2 |
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.