| Benchmark | GPT-5.6 Sol | o4-mini |
|---|---|---|
| AA Agentic Index | 57.8 | 36.1 |
| AA Intelligence | 61 | 26.1 |
| AA-LCR | 77.7 | 60 |
| AA-Omniscience | 22 | -35.7 |
| arc_agi_1 | 97.5 | 58.7 |
| ARC-AGI-2 | 92.5 | 6.1 |
| Artificial Analysis Coding Index | 78.3 | 25.6 |
| browsecomp | 90.4 | 51.5 |
| critpt | 32.3 | 0.6 |
| gdpval | 61.1 | 25.4 |
| GPQA Diamond | 94.6 | 81.4 |
| HLE | 49.5 | 16.5 |
| IFBench | 72.7 | 68.7 |
| MMMU-Pro | 84.6 | 69.2 |
| OmniScience Accuracy | 59.4 | 24.8 |
| OmniScience Non-Hallucination | 10.6 | 19.5 |
| scicode | 56.9 | 46.5 |
| simplebench | 64.8 | 38.7 |
| simpleqa_verified | 71.6 | 23.9 |
| strongreject | 98.5 | 96 |
| Terminal-Bench Hard | 65.9 | 15.2 |
| τ²-Bench Telecom (AA run) | 85.1 | 55.6 |
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.