GPT-Daybreak Released to Accelerate Defenders

OpenAI has released GPT-Daybreak for cybersecurity defenders. The strategy is to provide cutting-edge models to trusted defense organizations before attackers automate cyberattacks with AI.

Daybreak consists of two access tiers.

• Daybreak Blue: Provides general-purpose frontier models such as GPT-5.6 Sol, used for vulnerability discovery, secure code review, malware analysis, incident response, and patch verification.
• Daybreak Red: Provides specialized cybersecurity models for authorized vulnerability research, exploit verification, and security testing.

GPT-5.6-Cyber, provided in Daybreak Red, is based on GPT-5.6 Sol and has been further trained for specialized tasks such as zero-day vulnerability discovery and exploit chain development. It focuses on reducing unnecessary refusals for high-risk dual-use cyber tasks.

In OpenAI's internal evaluation, the Advanced Cybersecurity Completion Rate, GPT-5.6-Cyber's request completion rate was 95.0%. GPT-5.6 Sol recorded 1.5% with system safeguards applied and 2.0% in Daybreak Blue, while the previous model, GPT-5.5-Cyber, recorded 57.3%.

OpenAI also provided early access to trusted customer partners. Security research firm SpecterOps evaluated that GPT-5.6-Cyber more accurately tracks real-world exploit constraints and complex states, completing tasks in one day that previous models failed to solve for weeks.