OpenAI announces on November 4 Daybreak Red is now available in AWS GovCloud with GPT-5.6 Cyber
Daybreak Red gives approved customers access to GPT-5.6 Cyber through Amazon Bedrock in AWS GovCloud for authorized vulnerability research, exploit validation and security testing.
OpenAI announced on November 4 that its cyber defence service Daybreak Red is now available in the AWS GovCloud US-East region, giving approved customers access through Amazon Bedrock to GPT-5.6 Cyber, a model purpose-trained for cybersecurity, for authorized vulnerability research, exploit validation and security testing.
Access is aimed mainly at government security teams and verified cybersecurity defenders, and requires two separate steps: obtaining OpenAI Daybreak Red approval, and AWS provisioning the access. OpenAI states explicitly that Daybreak Red approval does not automatically enable model access on AWS, and customers still need to work with their AWS account team to activate it. Applicants can submit the Daybreak application form or contact their OpenAI and AWS account teams to confirm eligibility.
The page also notes that when Daybreak Red is used through Amazon Bedrock, inference data is not used to train models and customers do not need to share their Daybreak Red data with OpenAI. This launch extends frontier cybersecurity models from commercial cloud regions into a government compliance region, and continues the quarterly iteration cadence OpenAI has established for its cyber model line since the first cyber model in April 2026.