OpenAI GPT-5.4 and GPT-5.5 arrive on Amazon Bedrock through an OpenAI-compatible Responses API endpoint

OpenAI's GPT-5.4 and GPT-5.5 became available on Amazon Bedrock on June 1, 2026, through an OpenAI-compatible Responses API endpoint. The move lets AWS customers access frontier OpenAI models within their existing Bedrock environment—using familiar AWS authentication and billing—without a separate OpenAI API subscription.

What's new

Both of OpenAI's current production text models are now accessible through Amazon Bedrock:

• GPT-5.5 — $5/$30 per million input/output tokens, 1M context, 128k max output
• GPT-5.4 — $2.50/$15 per million input/output tokens, 1M context, 128k max output

The Bedrock endpoint mirrors OpenAI's Responses API shape, so existing code written against the OpenAI SDK can be redirected to the Bedrock endpoint without changing the request structure. AWS IAM authentication handles authorization, meaning enterprise customers with committed AWS spend can apply those credits toward OpenAI model inference.

Regional availability varies by model at launch.

Context

Amazon Bedrock has operated as a multi-provider model hub since 2023, initially featuring Anthropic, Cohere, and AI21 Labs. Anthropic's Claude expanded broadly on Bedrock in April 2026—becoming generally available across 27 AWS regions—and Anthropic followed with its own "Claude Platform on AWS" managed layer in May 2026.

Historically, OpenAI models were only accessible through OpenAI's own API or Azure OpenAI Service, Microsoft's exclusive cloud partner for OpenAI models. Adding Amazon as a distribution channel represents the first significant break from that two-channel model since OpenAI launched its API.

The move places OpenAI models in direct competition with Anthropic's Claude models in Bedrock's model selection interface—the same console where enterprise teams already run A/B tests, manage rate limits, and handle billing for AWS-hosted AI.

Why it matters

For enterprise teams that have standardized on Bedrock as their model abstraction layer, having GPT-5.4 and GPT-5.5 available in the same interface as Claude, Llama, and other models removes a meaningful friction point. Instead of managing separate OpenAI API credentials, spend tracking, and billing relationships, Bedrock-native teams can now compare OpenAI and Anthropic outputs in a unified environment with a single monthly invoice.

For OpenAI, Bedrock access could accelerate enterprise adoption considerably. Large AWS customers with negotiated committed-use discounts can now apply those credits to GPT-5.x inference—an economic incentive structure that Azure OpenAI Service has long offered exclusively to Microsoft's cloud customers.

The OpenAI-compatible Responses API shape is a strategic choice as well: it means OpenAI's API design is now a de-facto interface standard that Bedrock is explicitly adopting, rather than abstracting away. That increases switching costs out of the OpenAI API surface even for teams using Bedrock as their primary entry point.