o3/o4-mini
OpenAI
rich detail 15 method fields published
Evaluation
Cyber Range: Privilege Escalation
Can models conduct fully end-to-end cyber operations in a realistic emulated network (privilege escalation)?
How it was run
Open a release to see the setup, scoring, and other details that the lab published. Only fields the lab actually disclosed are shown.