On November 6, Beijing time, OpenAI announced the launch of GPT-6 Cyber, a cybersecurity model that has been trained to detect and verify exploitable vulnerabilities while never going beyond the scope set by the researchers. The first batch of GPT‑6 Cyber will be opened to a small number of approved Daybreak Red customers through APIs and codex, and will gradually expand the scope of access to all DayBreakRed customers, while also opening GPT‑5.6 Cyber to all Daybreak Red customers.

On November 6, Beijing time, OpenAI announced the launch of GPT-6 Cyber, a cybersecurity model that has been trained to detect and verify exploitable vulnerabilities while never going beyond the scope set by the researchers. The first batch of GPT‑6 Cyber will be opened to a small number of approved Daybreak Red customers through APIs and codex, and will gradually expand the scope of access to all DayBreakRed customers, while also opening GPT‑5.6 Cyber to all Daybreak Red customers.