OpenAI models on AWS Bedrock as Amazon adds Codex and Bedrock Managed Agents
Amazon adds OpenAI models, Codex and Bedrock Managed Agents to AWS after OpenAI revised its Microsoft deal on April 27–28, 2026, expanding cloud AI choice.
OpenAI models on AWS Bedrock have been made available to customers after Amazon announced on April 28, 2026 that its Bedrock service now supports OpenAI’s latest reasoning models and the Codex code-generation service. The move follows a revision to the OpenAI–Microsoft agreement announced on April 27, 2026 that removed exclusive rights and cleared the way for other cloud providers to offer OpenAI products. Amazon also introduced a new Bedrock Managed Agents product aimed at enterprises that want pre-configured agent capabilities built on OpenAI technology.
Amazon integrates OpenAI models into Bedrock
Amazon said the Bedrock catalog will now include the latest OpenAI reasoning models alongside multiple foundation models from other vendors. This integration lets developers choose OpenAI models directly within AWS’s managed model-hosting environment for application development and deployment.
The company also confirmed the inclusion of Codex, its code-generation toolset, enabling developers to run code-completion and developer-assistant workflows inside Bedrock. AWS framed the additions as expanding developer tooling and accelerating the build-out of AI-enabled applications on its platform.
Bedrock Managed Agents launched for enterprise workflows
The new Bedrock Managed Agents product is designed to create and operate multi-step autonomous workflows using OpenAI reasoning models. Amazon positioned the offering for enterprises seeking agent steering, observability, and built-in security controls without assembling agents from raw components.
According to Amazon, Managed Agents provide configurable policy controls, role-based access, and logging to help companies meet compliance and operational requirements. The service is targeted at customers who want managed agent behaviour and simplified lifecycle management for production deployments.
Timeline and implications of the revised Microsoft agreement
OpenAI’s revision to its agreement with Microsoft, disclosed on April 27, 2026, removed Microsoft’s exclusive product rights and clarified which clouds can commercially host OpenAI technology. The change followed OpenAI’s earlier multi-billion-dollar arrangement with Amazon that signaled a broader cloud strategy for the company.
Amazon’s announcement the next day underscored how quickly cloud providers can move to incorporate newly available models once contractual restrictions are adjusted. AWS CEO Andy Jassy also commented publicly on the development after the agreement revision, noting the change marked a significant commercial opening for other cloud vendors.
Shifting alliances across the cloud and AI industry
The Bedrock addition highlights a broader reshuffling of partnerships in generative AI, with major cloud and AI firms aligning around different model makers. Microsoft has deepened ties with other model developers, while OpenAI has expanded its commercial relationships beyond a single cloud partner.
These shifting alliances are prompting cloud vendors to accelerate product launches and differentiate through managed services, integration, and security features. For customers, the fragmentation of supply can offer more choice but also increases the complexity of comparing performance, pricing, and governance across providers.
What enterprises should expect from the new availability
With OpenAI models on AWS Bedrock, enterprises can expect more deployment options and potentially tighter integration with AWS services such as identity, monitoring, and data stores. The addition of Codex and managed agent features could shorten development cycles for AI-driven applications, particularly in developer tooling and automation scenarios.
However, organizations will need to evaluate data residency, model fine-tuning options, and compliance controls before migrating critical workloads. Customers weighing multi-cloud strategies should also consider vendor SLAs, cost models, and the operational overhead of supporting agents at scale.
Amazon’s inclusion of OpenAI technology on Bedrock marks a notable moment in cloud competition. By combining model access with managed infrastructure and agent orchestration, AWS seeks to offer an integrated path to production for enterprises adopting generative AI.
The market response over the coming months will show whether broader model availability drives faster enterprise adoption or mainly intensifies provider competition over ease of use, security, and total cost of ownership.
