benchgap
OpenAI · model

GPT-6 Astra (max) benchmark scores

As of 2026-10-07, GPT-6 Astra (max) (OpenAI) has measured scores on 13 benchmarks and estimated scores on 2 more.

BenchmarkScoreSource
Terminal-Bench 2.188.4%measured
Terminal-Bench 4.059.1%measured
Terminal-Bench Science 0.163.3%measured
GPQA Diamond96.1%measured
SciCode56.5%measured
CritPt31.7%measured
GDP.pdf31.0%measured
τ³-Bench Banking41.4%measured
MMMU-Pro87.0%measured
AA-AnalystAgent51.2%measured
Harvey LAB93.6%estimated ± 1.6 pp, low confidence
Humanity's Last Exam54.7%measured
AutomationBench68.5%measured
EnterpriseOps-Gym52.2%estimated ± 1.8 pp, low confidence
ITBench SRE48.6%measured