JA EN
shiichan

Bedrock in AWS GovCloud (US) Expands API Support and Adds Cross-Region Inference for OpenAI Models

AWS What's New ★★★★ View Markdown
#aws#ai#infrastructure

Hi everyone, it's Shii-chan! Big news for anyone using OpenAI models in GovCloud!

AWS What's New aws.amazon.com

What was announced?

AWS What's New announced that Amazon Bedrock expands support for OpenAI GPT-5.6 models in AWS GovCloud (US).

  • Models: GPT-5.6 Terra and Luna
  • APIs: Responses, Converse and Chat Completions (all on the bedrock-runtime endpoint)
  • New: Cross-Region Inference within AWS GovCloud (US)

What changes

Cross-Region Inference automatically routes requests across both AWS GovCloud (US) Regions. You get higher throughput and redundancy without managing capacity across Regions, while maintaining US data sovereignty and handling demand spikes.

Because the native OpenAI APIs now run on bedrock-runtime, the same account-level controls you use for other Bedrock models apply.

Dive Deep

You can track usage in:

  • Bedrock model invocation logging (deliverable to Amazon S3 or CloudWatch Logs)
  • CloudWatch metrics (invocation counts, token counts, latency, throttles, errors)
  • AWS Cost Explorer and the Cost and Usage Report (attribute spend by model)

For details, see the GPT-5.6 (Terra and Luna) model cards in the Amazon Bedrock User Guide.

Wrap-up

  • GPT-5.6 Terra / Luna are available on bedrock-runtime in GovCloud (US)
  • Responses, Converse and Chat Completions APIs are supported
  • Cross-Region Inference within GovCloud (US) gives higher throughput and redundancy
  • Logging, metrics and cost tracking work like other Bedrock models

A good fit for teams in government or regulated environments that want OpenAI models!

Related posts