OPENAI has launched GPT-5.6-Cyber, a large language model designed for cybersecurity tasks, alongside an updated two-tier security access program named Daybreak. The two tiers include Daybreak Blue for defensive tasks and Daybreak Red for advanced cybersecurity capabilities.
Members of Daybreak Blue can utilize frontier models like GPT-5.6 Sol with system-level safeguards for authorized security work, while Daybreak Red provides access to GPT-5.6-Cyber, which specializes in tasks such as vulnerability research and exploit validation. OpenAI claims GPT-5.6-Cyber significantly outperforms previous models in executing sensitive cybersecurity requests. Notably, it achieved a 95% completion rate on specific tasks compared to just 1.5% for general-access GPT-5.6 Sol. The update aims to mitigate risks and enhance cybersecurity workflows by allowing tailored access based on task requirements.