OpenAI GPT-5.5, GPT-5.4 and Codex Arrive on Amazon Bedrock

What happened?

On January 15, 2026, AWS announced the integration of OpenAI's GPT-5.5, GPT-5.4 models and the Codex coding agent into its Amazon Bedrock platform. According to the AWS News Blog, these models are now available to customers through the Responses API in Bedrock's next-generation inference engine, designed for high performance, reliability, and security.

Technical details and availability

GPT-5.5 and GPT-5.4 are multimodal and advanced reasoning models. According to OpenAI, GPT-5.5 is optimized for the most demanding workloads, while GPT-5.4 offers the best balance between price and performance. Both models can be invoked via the OpenAI SDK or directly with curl, targeting the bedrock-mantle endpoint. Codex, on the other hand, is a coding agent powered by GPT-5.5 that enables writing, refactoring, debugging, testing, and validating code in large repositories. Over 4 million developers use it weekly, according to OpenAI.

The integration ensures that all inference is processed within the selected Bedrock region, meeting data residency requirements. The pricing model is per token, with no per-seat licenses or per-developer commitments.

Why is this important?

This move represents a significant shift in the distribution strategy of AI models. Traditionally, OpenAI offered its models directly or through Azure. The arrival on Bedrock expands the ecosystem and offers enterprises an additional option with the advantages of AWS infrastructure: scalability, security, and regulatory compliance. Moreover, the integration with Codex as a coding agent embedded in IDEs like Visual Studio Code, JetBrains, and Xcode facilitates adoption in existing development workflows.

Market implications

This alliance could redefine competition in the enterprise AI model market. AWS directly competes with Azure OpenAI Service and Google Cloud Vertex AI. The availability of OpenAI's frontier models on Bedrock could accelerate generative AI adoption in companies already using AWS, reducing integration friction. Additionally, the pay-per-token model with no per-developer commitments democratizes access to advanced coding capabilities.

What readers should know

• Practical access: Developers can start using GPT-5.5 and GPT-5.4 on Bedrock by setting appropriate environment variables and using the OpenAI SDK or curl.
• Codex as an agent: Codex is not just a model but a full agent that can integrate into IDEs and CLI, facilitating complex software development tasks.
• Privacy and compliance: All processing stays within the selected Bedrock region, which is crucial for companies with data residency requirements.
• Pricing: No per-seat fees; you only pay per token consumed, allowing scaling based on actual usage.

According to AWS, “with GPT-5.5 powering inference, Codex introduces a new class of intelligence optimized for complex, long-horizon developer workflows.”

Historical context

The partnership between AWS and OpenAI is not new, but the depth of this integration is. In 2024, AWS began offering Anthropic models on Bedrock, and now with OpenAI it completes its frontier model offering. This reflects an open platform strategy where AWS does not limit itself to a single model provider but offers multiple options to its customers.

Speculation and unconfirmed

Although the announcement is official, no details have been revealed about financial agreements between AWS and OpenAI, nor about the exclusivity or duration of the deal. It has also not been confirmed whether other OpenAI models, such as DALL-E or Whisper, will come to Bedrock in the future.