Earlier in 2026, OpenAI's frontier models and Codex became generally available on AWS, providing enterprises with a new way to deploy advanced AI in production. Now, OpenAI has taken the next step in its collaboration with AWS by making Daybreak capabilities accessible through Amazon Bedrock.
Through Daybreak Access, security professionals can leverage frontier cyber models within their existing AWS environments. Both Daybreak Blue and Daybreak Red access levels are available on AWS:
- Daybreak Blue offers access to frontier general-purpose models, including GPT-5.6 Sol, with safeguards designed for authorized defensive security operations.
- Daybreak Red offers access to OpenAI's purpose-trained cybersecurity models for authorized vulnerability research, exploit validation, and security testing.
These models help speed up vulnerability research, detection engineering, and incident response-from initial discovery through to a validated fix. They also support complex workflows such as exploit reproduction and mitigation development.
Delivering Frontier Cybersecurity Capabilities via AWS
For enterprises, adopting specialized cybersecurity tools requires more than just model performance-it also demands security review, governance, procurement, access controls, and an operational model teams can sustain.
Through Amazon Bedrock, eligible customers can use Daybreak, including both Daybreak Red and Daybreak Blue, within the AWS environments where they already build, secure, and operate software. This provides security teams with a clearer path to leverage frontier AI through familiar AWS security, governance, and operational workflows.
Together, OpenAI and AWS are enabling more organizations to deploy advanced cybersecurity capabilities in production.
Getting Started
Daybreak Red and Daybreak Blue require enrollment in Daybreak Access. Once approved, the model can be accessed through the Amazon Bedrock console or the Responses API using the bedrock-mantle endpoint. For more details, refer to the documentation.