OpenAI and Amazon Web Services (AWS) have announced the general availability of OpenAI's frontier models through AWS. The latest GPT-5.5 and GPT-5.4 can now be called from Amazon Bedrock, the cloud AI platform, and the coding agent Codex can use the same models as well. The rollout began on June 1 local time and is limited to US regions for now, with availability in Japan still undecided.

Calling the Latest Models Directly from Amazon Bedrock

Amazon Bedrock is an AWS platform for working with a range of AI models through a single API. Until now it has centered on models from Anthropic, Meta, and Amazon itself, but this update formally adds OpenAI's frontier models to the lineup.

The available models are GPT-5.5 and GPT-5.4, which fit straight into use cases such as text generation, summarization, and code writing. In addition, OpenAI's coding agent Codex can reach the same models through Bedrock, giving developers a much wider set of options.

US Regions Only for Now, Japan Still Undecided

The rollout is staged. At announcement it covers US regions only: GPT-5.5 is available in US East (Ohio), and GPT-5.4 in US East (Ohio) and US West (Oregon).

For the Japan region, no availability date has been set. Teams that want to run these models in production from Japan will need to plan around which regions are supported. Coverage is likely to expand, so it is worth keeping an eye on each model's region availability list.

Authentication Handled with Bedrock API Keys or IAM

A key detail of this update is how authentication works. When using OpenAI models through Bedrock, you do not need a ChatGPT sign-in or an OpenAI API key. Instead, you use a Bedrock API key or AWS IAM, the AWS identity and access management service.

For companies and developers who already build on AWS, this is a real advantage. They can fold OpenAI's models into their familiar AWS permission design without managing a new external account or key. Billing and access control can also stay consolidated on the AWS side, which helps keep operational overhead down.

Some Limits When Used Through Codex

There are trade-offs as well. When Codex is used with a Bedrock API key, the response-speeding Fast Mode and cloud-based agent features are not available. Image generation and voice input are also unsupported, so for workflows that need those, sticking with an OpenAI API key remains the practical choice.

In other words, going through Bedrock is an option optimized for ease of integration into AWS environments, and it does not carry over every Codex feature as-is.

OpenAI Models Spreading Beyond Microsoft's Cloud

This release is also a milestone in OpenAI models becoming broadly usable outside Microsoft's cloud. OpenAI's models have long felt closely tied to Microsoft, but offering them formally from AWS, another major cloud, makes it easier for users to choose a cloud that fits their environment.

Being able to use the same model across cloud providers reduces lock-in to a single service and lets teams weigh price and performance when deciding where to deploy. Greater freedom in where AI models run is a change developers are likely to welcome.

Summary

OpenAI's frontier models are now generally available on Amazon Bedrock, putting GPT-5.5, GPT-5.4, and Codex within reach from the AWS platform. Availability is limited to US regions for now and undecided for Japan, but because authentication is handled entirely with Bedrock API keys or IAM, integrating into AWS environments is straightforward. While Codex via Bedrock lacks Fast Mode and image generation, the move to make OpenAI models available beyond Microsoft's cloud steadily expands the freedom to choose where to run them.