OpenAI models, Codex, and Managed Agents land on AWS
OpenAI and AWS announced an expanded strategic partnership today, bringing OpenAI’s capabilities directly into the AWS ecosystem across three fronts: OpenAI models on Amazon Bedrock, Codex on AWS, and the new Amazon Bedrock Managed Agents powered by OpenAI.
On the model side, OpenAI’s frontier models including GPT-5.5 will be available on Amazon Bedrock, letting enterprise customers call them through AWS’s security controls, identity systems, and procurement processes. Teams already running critical workloads on AWS no longer need to choose between cloud infrastructure and OpenAI’s API — they get a unified path from experimentation to production.
Codex, now used by over 4 million people weekly, can be configured to use Bedrock as its backend provider. Companies with AWS commitments and Bedrock access can run Codex through AWS billing, security, and high-availability infrastructure, with eligible usage counting toward cloud commitments. This reduces procurement friction significantly for enterprise adoption, starting with Codex CLI, the desktop app, and VS Code extension.
The most strategic piece is Amazon Bedrock Managed Agents powered by OpenAI. This lets enterprises build agents that maintain context, execute multi-step workflows, use tools, and act across complex business processes — all within their trusted AWS environment. AWS handles deployment, orchestration, and governance, so teams can focus on making agents useful rather than assembling infrastructure around them. This substantially lowers the bar from agent demo to production deployment.
For OpenAI, this is a major channel expansion. AWS serves the largest enterprise customer base globally, and distributing models and agent capabilities through Bedrock lets OpenAI reach organizations that prefer AWS procurement over direct API access. For AWS, adding GPT-5.5 and Codex to Bedrock strengthens its position as the enterprise AI gateway.
Notably, OpenAI restructured its Microsoft partnership earlier this year, giving Microsoft some model autonomy. This deep AWS tie signals a clear multi-cloud distribution strategy — let customers choose their cloud environment rather than locking them into a single provider.