OpenAI has introduced GPT-6 Cyber, a specialized cybersecurity model trained to find and fix vulnerabilities while staying within boundaries set by researchers. The model was released on November 5, 2026, and is already helping secure real-world software by finding and validating thousands of vulnerabilities in security-critical codebases such as Firefox and Linux.
GPT-6 Cyber Gains Across Cybersecurity Benchmarks
GPT-6 Cyber also showed major gains on cybersecurity benchmarks, achieving a 51% improvement on ExploitGym as compared with GPT-5.6 Cyber. It also aligned with security testing requirements, avoiding out-of-scope exploitation in 99.93% of tests without deployment prompts and 100% with prompts.
GPT-6 Cyber is a significant step up from GPT-5.6 Cyber and outperforms GPT-6 Astra on most cyber tasks. OpenAI calls the model Critical in cybersecurity under its Preparedness Framework, up from High for GPT-5.6 Cyber.
Real-World Results of GPT-6 Cyber
OpenAI’s vulnerability researchers use GPT-6 Cyber to reproduce 1,724 vulnerabilities and develop working exploits for 948, more than five times the 166 exploited by GPT-5.6 Sol using a similar research harness.
In a recent vulnerability research campaign, GPT -6 Cyber produced evidence and reproduced tests for 81 novel vulnerabilities classified by Mozilla as very high severity and showed an additional 30 known issues. Mozilla has started implementing fixes for the issues reported. The model also helped identify critical weaknesses in Brotli, a compression library used by Node.js and other software.
Beyond vulnerability discovery, security teams at Cisco, Calif and CodeRabbit use it for vulnerability research and code review. OpenAI has also used the model to identify and address weaknesses in its own infrastructure through controlled security testing. OpenAI’s ultimate goal is to expand Daybreak, bringing advanced security research to organizations of every size, independent researchers, and open-source maintainers.
Stronger Safeguards for Detecting Misuse
However, the models’ advanced capabilities can also create misuse risks. Thus, OpenAI has implemented stronger safeguards to detect and disrupt misuse and keep the models’ actions within the user’s intent and authorization.
This includes cyber-specific monitoring, protection against stolen credentials, misuse detection, sandboxing recommendations, and even restrictions for critical infrastructure environments. The company is working with cybersecurity organizations such as Dragos to evaluate risks that involve operational technology and industrial control systems.
Availability and What’s Ahead
GPT-6 Cyber is initially available through the API and Codex for a select group of approved Daybreak Red customers. The company is planning to expand access to all Daybreak Red customers soon. OpenAI’s new release highlights the role of advanced AI in cybersecurity, offering a stronger approach to identifying vulnerabilities faster, validating threats, and improving security over time.
Head over to HiTechNectar to read all the latest news around the technology landscape.
Also Read:
Meet OpenAI’s GPT-5.5: Features, Pricing, Benchmarks & Real-World Use Cases Explained
