OpenAI · model
GPT-5.6 Cyber benchmark scores
As of 2026-11-05, GPT-5.6 Cyber (OpenAI) has measured scores on 1 benchmark and estimated scores on 14 more.
| Benchmark | Score | Source |
|---|---|---|
| GDPval-AA | 55.4% | estimated ± 7.4 pp, low confidence |
| BrowseComp | 90.6% | estimated ± 3.0 pp, medium confidence |
| CyberGym | 86.3% | estimated ± 10.0 pp, medium confidence |
| AA Agentic Index | 49.6% | estimated ± 8.0 pp, low confidence |
| AA-AnalystAgent | 46.6% | estimated ± 4.6 pp, high confidence |
| MazeBench | 12.3% | estimated ± 3.0 pp, high confidence |
| AA AutomationBench | 70.5% | estimated ± 6.2 pp, medium confidence |
| DeepsecBench-AA | 37.0% | estimated ± 8.1 pp, low confidence |
| GDP.pdf | 27.8% | estimated ± 7.2 pp, low confidence |
| AA ITBench | 51.8% | estimated ± 4.3 pp, high confidence |
| CWE-bench v1 | 61.5% | estimated ± 4.5 pp, medium confidence |
| Hermes Bench | 73.8% | estimated ± 0.3 pp, medium confidence |
| Hermes Index | 53.0% | estimated ± 2.6 pp, medium confidence |
| ExploitGym | 32.1% | measured |
| Agents' Last Exam | 51.7% | estimated ± 9.6 pp, medium confidence |