OpenAI has announced that its Daybreak cybersecurity models are now available through Amazon Bedrock, according to announcements from both openai.com and aws.amazon.com published on August 11, 2026.
According to openai.com, the Daybreak offering includes two access levels:
- Daybreak Blue provides access to frontier general-purpose models, including GPT-5.6 Sol, with safeguards tailored for authorized defensive security work
- Daybreak Red provides access to purpose-trained cybersecurity models for authorized vulnerability research, exploit validation, and security testing
According to aws.amazon.com, Daybreak Red specifically provides access to GPT-5.6 Cyber, a purpose-trained cybersecurity model. The AWS announcement notes that “AWS security teams are using both models today to analyze source code, discover vulnerabilities, and conduct red-team research,” according to John Sheehan, Vice President of AWS Security.
Both sources emphasize that the models are designed to help accelerate vulnerability research, detection engineering, and incident response. According to openai.com, these capabilities support “complex workflows such as exploit reproduction and mitigation development.”
Access to both Daybreak Red and Daybreak Blue requires enrollment in Daybreak Access, according to openai.com. Once approved, eligible customers can access the models through the Amazon Bedrock console or the Responses API using the bedrock-mantle endpoint, operating within existing AWS security and governance frameworks.