The company introduced Daybreak in May to help ecosystem partners use its advanced AI models to adapt to evolving threats. The expansion includes Daybreak Blue, offering unique access to advanced general-purpose models with altered safeguards for defensive security work, and Daybreak Red, which provides access to purpose-trained cybersecurity models for security testing and vulnerability research. Daybreak Red users will also have access to the new GPT-5.6-Cyber model, built on GPT-5.6 Sol, designed to improve capabilities and reduce refusals on specialized cybersecurity tasks.