AWS Deepens OpenAI Ties, Expands Bedrock Access

Amazon Web Services (AWS) and OpenAI said they have broadened their partnership by bringing OpenAI’s newest models and its coding agent, Codex, into Amazon Bedrock. The companies said the integration is available in limited preview and is aimed at giving enterprise customers more advanced AI options inside AWS environments.

Under the expanded arrangement, AWS customers can use Amazon Bedrock APIs to reach OpenAI’s frontier models while continuing to work within Bedrock’s existing framework. AWS said the approach is designed to offer a single access layer with consistent security, governance, and cost controls.

OpenAI models join Bedrock’s multi-provider lineup

AWS said the Bedrock integration lets customers deploy OpenAI models alongside models from other providers already available through the service, including Anthropic, Meta, and Mistral. The companies positioned the update as a way for enterprises to use multiple model families through one set of interfaces and operational controls.

The announcement also introduces Amazon Bedrock Managed Agents powered by OpenAI. AWS said the managed agents are intended to help customers move from experimentation to production by simplifying how OpenAI-powered agents are deployed on AWS.

Codex arrives for enterprise software teams

OpenAI’s Codex, described as a coding agent used by more than 4 million individuals each week, is now available through Amazon Bedrock. The companies said this enables enterprise development teams to use AI-assisted software development capabilities within their AWS setups, rather than relying on separate tooling outside their cloud environment.

AWS and OpenAI framed Codex’s availability on Bedrock as part of a broader push to make advanced AI tools accessible through existing enterprise infrastructure. The limited preview status means availability and rollout details may vary as the service is tested with selected users.

Managed Agents target production requirements

AWS said Amazon Bedrock Managed Agents powered by OpenAI combines OpenAI’s models and agentic capabilities with AWS’s global infrastructure and security. The service is intended to address common production needs for AI agents, including persistent memory, encoded skills, identity enforcement, and selecting appropriate compute options for deployed applications.

The companies also said the new offerings allow customers to count OpenAI model usage toward existing AWS cloud commitments. They described this as a way to streamline procurement and strengthen financial governance for organizations that manage AI spending through established AWS purchasing structures.

Why the integration matters for enterprise adoption

By placing OpenAI models and agent tooling inside Amazon Bedrock, AWS is positioning the service as a centralized route for enterprises that want access to multiple AI providers while keeping security and governance controls consistent. OpenAI, meanwhile, gains another enterprise distribution channel tied to AWS infrastructure and procurement processes.

Key details that remain unclear include the timeline for broader availability beyond limited preview and which specific “latest models” are included at launch. AWS and OpenAI did not provide additional rollout dates in the announcement.