OpenAI says Daybreak Red is available in AWS GovCloud (US-East)

OpenAI posted on 4 November 2026 that GPT-5.6 Cyber is available to approved customers in AWS GovCloud (US-East) through Daybreak Red on Amazon Bedrock. The post says Daybreak Red is for authorized vulnerability research, exploit validation and security testing, and that qualified security practitioners should review findings and proposed fixes before implementation. It says verified defenders get greater latitude for that work, with fewer refusals, alongside stronger identity verification, monitoring and access controls. It says that for Daybreak Red on Amazon Bedrock, inference data is not used to train models, and customers do not need to share that data with OpenAI.

The post says access requires OpenAI Daybreak Red approval and AWS provisioning, and that Daybreak Red approval does not automatically enable model access on AWS. It does not state a price. An AWS what's-new item, posted 30 October 2026, says Daybreak Red is available to eligible customers on Amazon Bedrock in AWS GovCloud (US-East) and in US East (Ohio). That AWS item also says inference data is not used to train models, and that Daybreak Red runs with zero-operator access enforced at the chip. The Ohio region and the zero-operator sentence are on the AWS item, not in the OpenAI post that was opened.

What it means for business: this is availability of an existing cybersecurity model for approved customers, not a new model and not a published price. Approval from OpenAI does not, on OpenAI's own post, turn access on in AWS by itself. This describes the two posts. It is not advice on an authorization to test a system.