AWS Brings GPT-5.6 Terra and Luna In-Country to India

Amazon Bedrock adds India-only GPT-5.6 inference profiles

AWS announced on August 27, 2026 that Amazon Bedrock now supports OpenAI GPT-5.6 Terra and Luna with India geographic cross-Region inference. The new profiles are designed for workloads that need model inference to remain within India while still drawing capacity from more than one AWS Region.

The two profile IDs are in.openai.gpt-5.6-terra and in.openai.gpt-5.6-luna. AWS says requests using these profiles can route only between Asia Pacific (Mumbai), ap-south-1, and Asia Pacific (Hyderabad), ap-south-2.

This is a regional availability and data-processing-boundary update, not a new OpenAI model release. GPT-5.6 was already available on Amazon Bedrock; the material change is the addition of India geographic profiles for Terra and Luna.

The India profiles keep inference processing inside the country

AWS says prompts and generated results sent through the India profiles stay within India, although a request can move between the Mumbai and Hyderabad Regions. That gives organizations a way to scale inference across two Indian Regions while maintaining an India processing boundary.

AWS identifies financial services, healthcare and public-sector use cases among the workloads that may have local processing requirements. Organizations still need to evaluate their own legal, regulatory, contractual and security requirements; using an India inference profile should not be treated as automatic compliance with every data-residency rule.

AWS says Amazon Bedrock uses a zero-data-retention model by default for model inputs and outputs, while noting an exception: for certain models including GPT-5.6, content flagged by automated abuse-detection classifiers can be retained for offline abuse detection according to Bedrock's documented policy.

Terra and Luna keep their long-context and multimodal capabilities

AWS says both GPT-5.6 Terra and Luna on the India profiles support a 1-million-token context window, accept text and image input, and produce text output.

Developers can invoke the profiles through OpenAI-compatible Responses and Chat Completions APIs on the bedrock-runtime endpoint or through Amazon Bedrock's Converse APIs. AWS recommends bedrock-runtime for new applications.

Existing OpenAI SDK applications can point their client at a Bedrock runtime endpoint in Mumbai or Hyderabad and supply the India geographic inference profile as the model identifier. Bedrock supports standard AWS credentials and Bedrock API-key authentication for these calls.

Prompt caching, observability and Codex are supported

AWS says prompt caching works with the India profiles, allowing repeated long prompt prefixes to be cached while remaining under the India geographic boundary. It states that cached reads receive the documented Bedrock prompt-caching discount, subject to current service pricing and caching rules.

CloudWatch and CloudTrail provide monitoring for profile invocations, and CloudTrail's inference-region metadata can show whether Mumbai or Hyderabad processed a request. AWS also documents use of the India profiles with the Codex CLI through the Amazon Bedrock Runtime model provider.

India profiles differ from Bedrock's global inference profiles

Amazon Bedrock also provides global cross-Region profiles that can route supported models across commercial AWS Regions worldwide to maximize capacity. Those profiles are not equivalent to the India profiles.

For workloads with India-local processing requirements, AWS specifically directs customers toward the in. geographic profiles, because the global profiles may process requests outside India.

Why this matters for India's enterprise AI market

For organizations already standardized on AWS, the new profiles reduce a practical deployment tradeoff between accessing newer frontier models and keeping inference processing inside India. They also give developers a single geographic model identifier while AWS handles capacity routing between the country's two supported Regions.

The announcement does not mean GPT-5.6 Sol is included in the India-local profiles described by AWS: the India launch specifically names Terra and Luna. It also does not guarantee that every Indian compliance regime is satisfied merely by selecting an India geographic profile.