OpenAI unveiled GPT-6 Cyber on Nov. 5, 2026, as a model for finding and validating exploitable vulnerabilities and helping developers fix them. Initial access will be limited to approved Daybreak Red customers through the API and Codex. OpenAI said the model is designed to stay within authorized testing boundaries; access also has safeguards, including default IP allowlisting and separate restrictions for critical-infrastructure work.
OpenAI said GPT-6 Cyber helped identify 81 previously unknown Firefox vulnerabilities that Mozilla classified as critical. Researchers shared findings and proposed fixes with Mozilla, which began rolling out updates. In Linux kernel research, the model developed 948 working exploits, OpenAI said. It scored 48.5% on ExploitGym, compared with 32.1% for GPT-5.6 Cyber.
