OpenAI introduced GPT-6 Cyber for vulnerability discovery, validation and security research. The company says initial access is through the API and Codex for selected approved Daybreak Red customers, with broader Daybreak Red access planned. The announcement describes scope-focused training and safeguards including misuse monitoring and default IP allowlisting for Daybreak Red API access. This entry reports the announced rollout and restrictions; it does not independently validate benchmark scores or claim universal availability.
Official source: OpenAI — https://openai.com/index/introducing-gpt-6-cyber/ Original publication date: 2026-11-05. Stored at midnight UTC to represent the calendar date, not a verified release time. Checked November 6, 2026.