OpenAI added GPT-6-Cyber to its API on November 5, a security-focused model that its developer changelog says can be used to "investigate vulnerabilities, reproduce them in an authorized test environment, and verify fixes." Access is restricted to organizations approved for OpenAI's Daybreak Red program.
According to OpenAI's model documentation, the model ID is gpt-6-cyber. It has a 1,050,000-token context window with up to 922,000 input tokens and 128,000 output tokens, accepts text and images, and supports reasoning tokens, streaming, structured outputs and function calling. It is available only through the Chat Completions and Responses endpoints.
Pricing is $20 per million input tokens, $2 per million cached input tokens and $100 per million output tokens, with cache writes at $25 for 30 minutes or $50 for 12 hours. In the Responses API the model can use web search, file search, code interpreter, a hosted shell, apply-patch, computer use and MCP tools.
Organizations must complete Daybreak setup before their first request, and OpenAI directs customers to their account teams to confirm eligibility. The gating reflects how labs are handling models with strong offensive-security capability: releasing them to vetted defenders through controlled programs rather than to the general API.
For security teams already in the Daybreak program, GPT-6-Cyber adds a model built for vulnerability reproduction and fix verification with very long context, which suits whole-codebase review. Teams outside the program cannot use it yet.