OpenAI Launches GPT-6-Cyber for Vulnerability Research and Patch Verification

01 What happened

OpenAI released GPT-6-Cyber, a model designed to investigate vulnerabilities, reproduce them in authorized test environments, and verify fixes. Use requires organization-level Daybreak Red access and approval.

02 Key details

• GPT-6-Cyber is designed to investigate vulnerabilities, reproduce them in authorized test environments, and verify fixes.
• Access requires organization-level Daybreak Red clearance and approval; the model is not publicly available.
• Pricing: $20 per 1M input tokens, $100 per 1M output tokens; 1,050,000-token context window with up to 128,000 output tokens.
• Supported tools via the Responses API include Web search, File search, Code interpreter, Hosted shell, Apply patch, and more.

03 Why it matters

The model provides cybersecurity professionals with a dedicated tool for vulnerability analysis and patch verification. Pricing: $20 per 1M input tokens, $100 per 1M output tokens; 1,050,000-token context window with up to 128,000 output tokens.

04 Who it matters to

Information security specialists.