| Benchmark | GPT-5.5 | o1 |
|---|---|---|
| AA Agentic Index | 47.4 | 31.1 |
| AA Intelligence | 56.3 | 24 |
| AA-LCR | 79 | 63.3 |
| AA-Omniscience | 20.5 | -10.6 |
| arena_vision | 1283 | 1193 |
| Artificial Analysis Coding Index | 74.9 | 39.7 |
| critpt | 27.1 | 0.3 |
| Fortress | 16.3 | 19.4 |
| gdpval | 49.5 | 11.5 |
| GPQA Diamond | 93.6 | 75.7 |
| HLE | 52.2 | 7.7 |
| IFBench | 75.9 | 70.3 |
| livebench | 80.7 | 52.3 |
| OmniScience Accuracy | 58 | 34.7 |
| OmniScience Non-Hallucination | 12 | 30.7 |
| scicode | 56.1 | 35.8 |
| simplebench | 69 | 41.7 |
| SWE-bench Verified | 80.6 | 48.9 |
| Terminal-Bench Hard | 60.6 | 12.9 |
| τ²-Bench Telecom (AA run) | 93.9 | 62.6 |
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.