OpenAI is expanding its OpenAI Daybreak program with two new access tiers—Daybreak Blue and Daybreak Red—to provide approved defenders with advanced cybersecurity capabilities using the newly introduced GPT-5.6-Cyber model.

  • Daybreak Blue grants access to frontier general-purpose models like GPT-5.6 Sol with tailored safeguards for defensive tasks such as vulnerability discovery and incident response.
  • Daybreak Red provides access to purpose-trained cybersecurity models, specifically GPT-5.6-Cyber, designed for authorized vulnerability research and exploit validation.
  • GPT-5.6-Cyber reduces refusals for high-risk dual-use cyber tasks, completing 95.0% of advanced scenarios compared to 1.5% for standard GPT-5.6 Sol.
  • The model outperforms previous versions on benchmarks like ExploitGym and has helped identify critical vulnerabilities in V8, mobile operating systems, and databases.

This expansion aims to democratize access to frontier intelligence for trusted defenders, helping them accelerate defensive workflows and prepare for AI-driven threats before attackers deploy offensive capabilities at scale.