OpenAI makes GPT-5.6 Cyber available in AWS GovCloud
OpenAI has made GPT-5.6 Cyber available to approved customers in AWS GovCloud (US-East) through its Daybreak Red service. The expansion gives eligible government security teams a way to use the model for authorized vulnerability research, exploit validation, and security testing within AWS environments.
- GPT-5.6 Cyber is available to approved customers in AWS GovCloud (US-East).
- Access is provided through Daybreak Red and Amazon Bedrock.
- Use cases include vulnerability research, exploit validation, and security testing.
- Inference data is not used to train models on Amazon Bedrock.
- Customers need both Daybreak Red approval and AWS provisioning.
Cybersecurity work for government teams
Daybreak Red provides access to OpenAI's purpose-trained cybersecurity models. According to OpenAI, GPT-5.6 Cyber can help defenders investigate complex software weaknesses, reproduce vulnerabilities in controlled settings, and develop mitigations.
The model is intended to support security work connected to critical missions. Teams can use it to reason through security findings, assess their impact, and evaluate proposed fixes. OpenAI says qualified security practitioners should review findings and proposed fixes before implementation.
Access through Amazon Bedrock
The availability in AWS GovCloud (US-East) builds on the availability of Daybreak models on Amazon Bedrock. It gives eligible government organizations a path to incorporate GPT-5.6 Cyber into workflows that operate within their AWS environments.
OpenAI says government organizations need to meet established security, governance, and operational requirements when adopting advanced cybersecurity capabilities. Daybreak Red is designed for verified defenders conducting authorized cybersecurity work, with fewer refusals for these tasks alongside stronger identity verification, monitoring, and access controls.
For Daybreak Red on Amazon Bedrock, inference data is not used to train models, according to OpenAI. Customers also do not need to share their Daybreak Red data with OpenAI.
Approval and AWS provisioning are required
Access requires both OpenAI Daybreak Red approval and AWS provisioning. Qualified security practitioners can begin by completing the Daybreak application or contacting their OpenAI or AWS account team about eligibility.
After approval, customers must work with their AWS account team to enable access. OpenAI says Daybreak Red approval does not automatically enable model access on AWS.
For government security teams, the announcement connects GPT-5.6 Cyber to an environment intended for eligible workloads rather than presenting access as generally available. Its practical value will depend on approval, provisioning, and the team's review of the model's findings and proposed mitigations before they are implemented.
OpenAI: Daybreak Red is now available in AWS GovCloud (US-East)
