Compiled by Futu News
AWS and OpenAI are introducing the latest OpenAI models to Amazon Bedrock, launching Codex on Amazon Bedrock, and initiating OpenAI-powered Amazon Bedrock Managed Agents (all in limited preview), providing enterprises with cutting-edge intelligence on trusted infrastructure.
Today, AWS announced a significant expansion of its partnership with OpenAI, bringing cutting-edge AI to the infrastructure that millions of organizations already trust. Enterprises aim to build using the most capable AI models and agents currently available, while also requiring the security posture, operational maturity, and data governance necessary for production-grade workloads. Starting today, we are combining these two elements through three new products (all in limited preview):
OpenAI Models on Amazon Bedrock: The latest OpenAI models, which customers can access, fine-tune, and orchestrate via the Amazon Bedrock API and console they already use.
Codex on Amazon Bedrock: An OpenAI programming agent on Amazon Bedrock designed for large-scale enterprise-grade software development.
OpenAI-Powered Amazon Bedrock Managed Agents: An optimized experience for building production-ready AI agents using OpenAI's leading models on AWS.
Collectively, these launches provide customers with the freedom and flexibility to select and use the best models for their use cases on the most widely adopted cloud platform.
OpenAI Models on Amazon Bedrock
Amazon Bedrock was built on the principle that customers should choose the best model for each use case. We are excited to announce that, starting today, the latest OpenAI models will be available on Amazon Bedrock. For the first time, AWS customers will be able to utilize OpenAI's cutting-edge models through the same services they already use for model access, fine-tuning, and orchestration. This means customers can evaluate and deploy OpenAI models alongside models from Anthropic, Meta, Mistral, Cohere, Amazon, and other leading providers under a single, consistent service with unified security, governance, and cost controls.
The OpenAI models on Bedrock inherit the full suite of enterprise-grade controls that customers already rely on: IAM-based access management, AWS PrivateLink connectivity, guardrails, encryption at rest and in transit, comprehensive logging via AWS CloudTrail, and integration with existing compliance frameworks. There is no need to configure additional infrastructure or learn new security models. Customers can account for the usage of OpenAI models within their existing AWS cloud commitments and consolidate AI spending with their broader AWS workloads. This simplifies procurement and financial governance for enterprises managing significant cloud investments on AWS.
Codex on Amazon Bedrock
Codex is rapidly becoming one of the best examples of how AI agents operate within enterprises. Over 4 million people use Codex weekly to automate coding tasks, write and refactor code, interpret complex systems, generate tests, and accelerate software delivery. With Codex on Amazon Bedrock, enterprise teams can utilize this OpenAI programming agent within the AWS environments they already operate at scale. Customers can authenticate using their AWS credentials, process inference through Amazon Bedrock infrastructure, and account for Codex usage within their AWS cloud commitments. Codex on Bedrock will initially support the Codex command-line interface, Codex desktop application, and Visual Studio Code extension via the Bedrock API.
Amazon Bedrock Managed Agents Powered by OpenAI
The most capable AI agents today have demonstrated what cutting-edge reasoning models can achieve: performing complex, multi-step tasks with minimal human intervention. OpenAI's leading models and agent capabilities represent the forefront of what is possible. However, production-grade AI applications require not only intelligence but also enterprise-level infrastructure, security, and operational foundations to run reliably at scale. Additionally, they need memory that persists across sessions, skills for coding procedures, identities with appropriate permissions, and compute options suited to the task. Currently, teams must build and assemble these components, which can be highly complex. To address these needs, we are excited to introduce Amazon Bedrock Managed Agents powered by OpenAI. Bedrock Managed Agents combine OpenAI's advanced models and agent capabilities with AWS's global infrastructure, security, and extensive services relied upon by millions of organizations. Using Bedrock Managed Agents, deploying production-ready, OpenAI-powered agents on AWS becomes fast and straightforward, allowing you to focus on what your agents should do rather than the underlying infrastructure.
Bedrock Managed Agents are optimized for OpenAI models on AWS. Built on the OpenAI Agent Framework (agent harness), they aim to unlock the full potential of OpenAI's cutting-edge models, enabling faster execution, sharper reasoning, and reliable guidance for long-running tasks. Security and governance are built in from the moment you deploy: each agent has its own identity, logs every operation for auditing, and operates within your environment, with all model reasoning performed on Amazon Bedrock. As customers scale to hundreds of thousands of agents within their enterprises, they will benefit from AWS's scalable global infrastructure and its proximity to the data, applications, and services they already rely on.
Box is a leading intelligent content management platform, helping over 115,000 organizations leverage enterprise AI to drive collaboration, manage their entire content lifecycle, protect critical content, and transform business workflows. 'Enterprises are currently seeking to deploy agents to provide solutions that bring their organizations into the next phase of AI,' said Ben Kus, Chief Technology Officer of Box. 'With Amazon Bedrock Managed Agents powered by OpenAI, developers can build optimized, production-scale AI applications that combine the advantages and capabilities of OpenAI's latest models with AWS's scale, security, and infrastructure. This combination will produce agents that continuously learn effective methods over time, tailor responses to the specific context of each user, and operate with the governance and auditability required by enterprises, all running on the cloud we already trust.'
Bedrock Managed Agents vs. Bedrock AgentCore
Bedrock AgentCore is an open platform for building, connecting, and optimizing large-scale agents using any model and framework. Bedrock Managed Agents are optimized for constructing agent solutions using OpenAI's cutting-edge models and agent capabilities. If you are building with Bedrock Managed Agents, AgentCore is a natural complement. AgentCore provides the default compute environment for Bedrock Managed Agents, and as your agents scale within the enterprise, AgentCore and Bedrock Managed Agents will offer additional capabilities such as authorization policy enforcement, agent and tool discovery, and observability and evaluation capabilities.
Future Outlook
This is just the beginning of the deepening collaboration between AWS and OpenAI. As OpenAI continues to push the frontiers of reasoning and agent capabilities, AWS and OpenAI will continue to bring the latest advancements to Amazon Bedrock—so the models and agents you use to build today will continue to benefit from new breakthroughs as they emerge.
Editor/Rocky