AWS · 2026-04-28 · major
OpenAI Models and Codex on Amazon Bedrock Reach General Availability — GPT-5.5, GPT-5.4, and Codex Now GA After April Preview
GPT-5.5, GPT-5.4, and Codex are now generally available on Amazon Bedrock. Pricing matches OpenAI's first-party rates with no AWS markup, Codex bills per token (no seat licenses), and inference runs on Bedrock's next-gen engine with full AWS governance (IAM, VPC, PrivateLink, KMS, CloudTrail).
OpenAI's frontier models reach a second hyperscaler less than 24 hours after the Microsoft exclusivity ended.
Key specs
| Ga models | GPT-5.5, GPT-5.4, Codex |
|---|---|
| Preview date | April 28, 2026 |
| Ga date | June 1, 2026 |
| Aws commitment | $35B financing line |
| Open ai commitment | 2 GW Trainium accelerators |
What is it?
Three new offerings on Amazon Bedrock, all in limited preview: the latest OpenAI models, Codex (the coding agent) accessible via Bedrock APIs plus the Codex CLI / desktop app / VS Code extension, and Bedrock Managed Agents — production-ready agents powered by OpenAI's frontier models and agent harness.
How does it work?
Customers call OpenAI models through Bedrock's existing inference endpoints with AWS IAM, PrivateLink, guardrails, encryption, and CloudTrail logging. Codex authenticates with AWS credentials and bills against existing AWS commits. Managed Agents adds persistent session memory, skill encoding, identity-based permissions, and audit logging on top of OpenAI's agent harness.
Why does it matter?
OpenAI was effectively Azure-exclusive for years; Microsoft's revised partnership last week cleared the path. AWS-native enterprises can now consolidate OpenAI usage onto an existing AWS commit instead of standing up a parallel Azure relationship, and OpenAI in turn gets a financing line tied to deploying 2 GW of Amazon Trainium accelerators.
Who is it for?
AWS-anchored enterprise teams and Codex users without an Azure footprint.
Try it
Register at https://aws.amazon.com/bedrock/openai