Run OpenAI models and Codex in AWS accounts and regions

OpenAI has made its frontier models and Codex generally available on Amazon Web Services, effective immediately, allowing enterprises to access and run those models from within AWS accounts and regions. The availability gives customers a path to use OpenAI model endpoints while keeping traffic, logs and controls inside AWS infrastructure.

OpenAI and AWS say the offering integrates model access into AWS account boundaries and management tooling. Customers can provision model access through their existing AWS identity and access management, network controls, and logging services. The release targets enterprise developers and security teams that require regional deployments, consolidated billing, and compliance alignment with existing cloud policies.

How the AWS integration works

OpenAI models are reachable via endpoints that operate inside a customer's AWS environment. Requests originate from applications running in the customer's VPC or AWS account and are routed to OpenAI-hosted model services while remaining subject to AWS network controls, identity policies and observability. Typical integration points include:

• AWS Identity and Access Management for controlling who can call model endpoints.
• VPC and networking controls to restrict traffic paths and apply firewall rules.
• Centralized logging with AWS services such as CloudTrail and CloudWatch to capture request metadata and operational events.
• AWS billing consolidation so model consumption appears on the customer's AWS invoice alongside other cloud costs.

The announcement emphasizes that enterprises keep account-level controls, while OpenAI continues to operate and update the underlying model infrastructure. The arrangement is pitched as a way to reduce integration effort for teams that already standardize on AWS tooling.

What developers and customers should know

Codex, OpenAI's model family for code generation and developer tooling, is included in the offering. That inclusion is designed to let engineering teams use code-completion and generation capabilities inside AWS-based development pipelines and CI systems. Developers can call the same generation and editing endpoints, while security and compliance teams monitor and manage usage through AWS consoles.

The integration does not remove OpenAI's operational role. Model updates, improvements and hosting remain under OpenAI's control. Customers should evaluate data handling, retention and privacy terms to confirm they meet internal and regulatory requirements before routing sensitive queries through the service. Enterprises with strict data residency or offline requirements will need to review the exact deployment model for regional availability and records handling.

Support and commercial terms are expected to follow standard enterprise channels. Organizations already using AWS Marketplace for software procurement will see familiar procurement and billing options. The integration may simplify procurement and cost accounting for teams that prefer to centralize vendor management and cloud spend in a single provider.

Why it matters

Making frontier models and Codex available inside AWS brings model access into standard enterprise controls and billing workflows, lowering a practical barrier for large organizations to adopt generative AI. Security, compliance and finance teams gain clearer paths to govern model use while developers can call advanced models from existing AWS-hosted applications and pipelines.