OpenAI · model
GPT-6 Astra (xhigh) benchmark scores
As of 2026-10-07, GPT-6 Astra (xhigh) (OpenAI) has measured scores on 6 benchmarks and estimated scores on 7 more.
| Benchmark | Score | Source |
|---|---|---|
| Terminal-Bench 2.1 | 88.2% | estimated ± 4.0 pp, medium confidence |
| Terminal-Bench 4.0 | 59.6% | measured |
| Terminal-Bench Science 0.1 | 57.9% | estimated ± 6.5 pp, medium confidence |
| GPQA Diamond | 96.3% | measured |
| CritPt | 31.4% | measured |
| GDP.pdf | 32.2% | measured |
| τ³-Bench Banking | 47.4% | estimated ± 5.0 pp, medium confidence |
| MMMU-Pro | 86.0% | measured |
| AA-AnalystAgent | 53.8% | estimated ± 7.8 pp, medium confidence |
| Harvey LAB | 94.0% | estimated ± 3.2 pp, high confidence |
| Humanity's Last Exam | 56.8% | estimated ± 3.4 pp, high confidence |
| AutomationBench | 67.2% | measured |
| EnterpriseOps-Gym | 42.4% | estimated ± 6.2 pp, low confidence |