GPT-5.5, GPT-5.4, and Codex are now generally available on Amazon Bedrock. Deploy them today in production applications and workers on Bedrock’s high-performance inference engine.
Key Takeaways
- GPT-5.5, OpenAI’s most advanced frontier model, is generally available on Amazon Bedrock. The price corresponds to OpenAI’s proprietary prices.
- The Codex on Amazon Bedrock is generally available with pay-per-token pricing. Inference is done through Bedrock and usage is factored into your existing AWS commitments.
In a significant development following our expanded partnership announcement, GPT-5.5, GPT-5.4, and Codex are now generally available on Amazon Bedrock, offering access to frontier models and the OpenAI coding agent for software development.
Amazon Bedrock is the platform for building and running production-scale AI applications and agents. OpenAI models on Bedrock run on Amazon Bedrock’s next-generation inference engine, designed for high performance, reliability, and security.
The Best Performing OpenAI Model on Amazon Bedrock
GPT-5.5 excels at understanding user intent and managing multi-step tasks autonomously. It is proficient in writing and debugging code across large codebases, analyzing data, generating documents and spreadsheets, and operating software across multiple tools until a task is complete. This model shows significant improvements in agent coding and knowledge work, crucial for maintaining context and action over time.
Both GPT-5.5 and GPT-5.4 are designed for complex, multi-step tasks and are available today in the Amazon Bedrock template catalog. You can access them through the Responses API on Amazon Bedrock, paying the same rate per token as directly from OpenAI, with no additional fees.
Bedrock’s Inference Engine
Bedrock’s inference engine provides your own isolated queue with automated capacity management, ensuring predictable performance even under heavy load. Each query’s complete state is captured durably and continuously, so if hardware fails or a node restarts mid-call, your query resumes from where it left off. Bedrock extends existing AWS governance controls like IAM, VPC, PrivateLink isolation, KMS encryption, and AWS CloudTrail audit logging. Your prompts and responses are not used to train models and are not shared with model providers, with these protections extending to GPT-5.5 and GPT-5.4 on Amazon Bedrock.
“At Amgen, we are focused on applying advanced AI in ways that accelerate the delivery of potential new therapies while equipping our teams with advanced tools. OpenAI’s GPT-5.5 and frontier models deliver compelling advances in capacity, quality, and consistency that matter in a field where questions are complex and standards for scientific accuracy and decision quality are exceptionally high. Making these models available on AWS gives us an important new avenue to explore and evolve these capabilities within responsible AI, including security, governance, and operational frameworks across the enterprise.” – Sean Bruich, Senior Vice President and Chief Technology Officer at Amgen.
Accelerate Software Development with Codex on Amazon Bedrock
Codex, the OpenAI coding agent for AI-based software development, is used by over 5 million people every week to write, refactor, debug, test, and commit code across large codebases. Codex maintains context across entire repositories, explains ambiguous failures, checks assumptions using tools, and applies changes to surrounding code considering how systems connect. Powered by GPT-5.5, Codex performs these tasks more efficiently and with higher quality compared to previous model releases.
Amazon Bedrock Codex is available through the Codex app, Codex CLI, and IDE integrations with Visual Studio Code, JetBrains, and Xcode, with all model inference routed through Amazon Bedrock. Inference stays within the selected region to meet data residency requirements. Payment is per token, with no seat licenses and no commitments per developer, allowing for quick start and scalable access over time.
“Autodesk is the technology platform for people who design and create the world around us. Workflows like building design are highly iterative, requiring precision, coordination, and continuous refinement across teams. With OpenAI and Codex models now generally available on Amazon Bedrock, our teams are evaluating how cutting-edge AI capabilities and AI-driven development tools on scalable and secure AWS infrastructure can help accelerate development workflows and support more enlightened decision-making for our customers.” – Ritesh Bansal, VP of Data Analytics, Agentic AI and AI/ML Platform at Autodesk.
What’s Next
When announcing our expanded partnership, we introduced Amazon Bedrock Managed Agents, powered by OpenAI. Coming soon, this will allow you to deploy production-ready agents based on the OpenAI Agent Harness, providing faster execution, more precise reasoning, and reliable steering of long-running tasks. Each agent will operate with its own identity, log every action for auditing purposes, and run all model inferences on Amazon Bedrock. To stay informed, register via the interest form.
We will continue to expand the OpenAI capabilities available on Amazon Bedrock as new advancements arrive. This includes Daybreak, OpenAI’s vision for transforming software development and defense. Daybreak, which includes cyber templates and Codex Security, is designed to help cyber defenders identify vulnerabilities, examine code for risks, and guide remediation throughout the development lifecycle. When Daybreak becomes available on Bedrock, security teams will be able to adopt it through the governance and operational frameworks they already use on AWS.
To Start
GPT-5.5 and GPT-5.4 are available today on Amazon Bedrock via the Responses API. Check the AWS Regions Page for availability. For documentation and a walkthrough, see Amazon Bedrock Documentation and the blog to get started.
About the Author
Bharat Sandhu
Bharat Sandhu leads AI/ML marketing for Amazon Web Services, covering silicon, models, training, inference, and agents. His team’s mission is to help customers build, deploy, and scale AI applications and agents faster, more securely, and at lower cost.
For more information, visit the original source here.
“`
