What changed
OpenAI moved GPT-6-Cyber to general availability on November 5, 2026. The release notes describe it as a model for authorized vulnerability research, including investigating flaws, reproducing them in a test environment, and verifying fixes.
Why it matters
This is a concrete productization of security-focused model use. For defenders, that can mean faster triage loops, easier lab reproduction, and more consistent patch validation inside approved environments.
Availability and takeaway
The model is intended for authorized users only and is surfaced through OpenAI’s platform docs and API tooling. The practical takeaway for developers and security teams is that a specialized workflow model is now available for sanctioned research, rather than requiring a generic model adapted by prompt alone.