Last updated: November 6, 2026.
OpenAI released GPT-6 Cyber on November 5, 2026 as a cybersecurity-specialized variant of GPT-6 Astra for authorized security research. The model targets vulnerability discovery, exploit development, web penetration testing and other advanced defensive workflows, but it is not a normal public ChatGPT model.
Quick answer: GPT-6 Cyber is OpenAI’s newest cyber-specialized model. It improves substantially over GPT-5.6 Cyber on exploit and vulnerability benchmarks and is initially available through Daybreak Red to approved users.
What is GPT-6 Cyber?
GPT-6 Cyber is a specialized version of GPT-6 Astra with additional cybersecurity post-training. It is designed for approved security teams that need deeper vulnerability research while staying inside the scope of an authorized engagement.
OpenAI lists use cases such as vulnerability discovery, exploit development, web pentesting, code review and validation of security fixes. The key difference from a general coding assistant is the combination of stronger cyber capability and a tightly controlled deployment model.
GPT-6 Cyber benchmarks
OpenAI reports a large jump over GPT-5.6 Cyber on ExploitGym. GPT-6 Cyber reached 48.5%, compared with 32.1% for GPT-5.6 Cyber.
On a web pentesting evaluation focused on discovering vulnerabilities without source-code access, GPT-6 Cyber reached 40.3% vulnerability recall, compared with 32.0% for GPT-6 Astra.
GPT-6 Cyber vs GPT-6 Astra
GPT-6 Astra remains OpenAI’s general frontier model for coding, science, computer use and professional work. GPT-6 Cyber goes further for specialized security work through cyber-focused post-training and a restricted deployment environment.
What is Daybreak Red?
Daybreak Red is OpenAI’s restricted-access program for advanced authorized cybersecurity work. Initial GPT-6 Cyber access is through the API and Codex for approved Daybreak Red customers.
The GPT-6 Cyber system card describes identity verification and organizational controls including single sign-on, multi-factor authentication and API IP allowlisting.
Can you use GPT-6 Cyber today?
Not as a standard ChatGPT model. OpenAI says it is initially available to a select group of approved Daybreak Red customers, with broader Daybreak Red access planned progressively.
Why this release matters
OpenAI classifies GPT-6 Cyber as Critical in cybersecurity under its Preparedness Framework. The company also says the model has already helped uncover and validate large numbers of exploits in security-critical codebases, including Linux and Firefox.
FAQ
Is GPT-6 Cyber available in ChatGPT?
No. Current access is described through Daybreak Red, the API and Codex for approved users.
Is GPT-6 Cyber stronger than GPT-6 Astra for security?
OpenAI reports that GPT-6 Cyber outperforms GPT-6 Astra on most cyber tasks in its evaluations.
Is GPT-6 Cyber public?
No. Access is restricted.
Sources
Explore more: AI Models – Latest AI Tools – All AI Tools
