# Amazon Bedrock Adds OpenAI GPT-5.6 Models with India-Only Inference > Source: > Published: 2026-08-19 01:15:59+00:00 **August 19, 2026**, (Inside AI) — Amazon Bedrock has expanded its model catalog to include **OpenAI GPT-5.6** models, **Terra** and **Luna**, with in-country inference support for **India**. The service now offers India Geo cross-Region inference profiles, enabling customers with data residency requirements to process inference requests entirely within Indian borders. The new profiles, **in.openai.gpt-5.6-terra** and **in.openai.gpt-5.6-luna**, route traffic exclusively across **Asia Pacific (Mumbai)** and **Asia Pacific (Hyderabad)** AWS Regions. This setup allows organizations to scale throughput while maintaining compliance with local data handling mandates. The models operate on the **bedrock-runtime** endpoint and support the **Responses**, **Chat Completions**, and **Converse** APIs. This move intensifies the competitive landscape for enterprise AI services in India. AWS already hosts models from **Anthropic**, **Meta**, and **Mistral** on Bedrock. Adding OpenAI models directly addresses demand from regulated sectors like banking, healthcare, and government, where data cannot leave the country. It also mirrors a broader industry pattern: cloud providers are racing to localize frontier model inference as national data laws tighten. Cross-Region inference is not new to Bedrock. AWS introduced the feature in **2024** to help customers avoid capacity bottlenecks by automatically routing requests across multiple regions. But the India Geo profile is a specific configuration that restricts routing to Indian infrastructure only. This distinction is critical for organizations that must prove data never transits through foreign servers, a requirement under India's evolving data protection framework. Customers using the new profiles retain access to existing Bedrock governance tools. Model invocation logs can be delivered to **Amazon S3** or **Amazon CloudWatch Logs**. Metrics flow through **Amazon CloudWatch**, and costs appear in **AWS Cost Explorer** and the **AWS Cost and Usage Report**. This consistency means enterprises can adopt OpenAI models without overhauling their compliance or financial operations workflows. ## Why India-Specific Inference Changes the Enterprise Calculus Data residency has shifted from a technical checkbox to a board-level risk issue. India's **Digital Personal Data Protection Act, 2023** imposes strict rules on cross-border data transfers for certain categories of personal data. While the law allows transfers to notified countries, many enterprises prefer to avoid ambiguity by keeping inference local. AWS's India Geo profiles remove that ambiguity for OpenAI models. The practical benefit is twofold. First, latency improves because requests stay within Indian data centers, avoiding round trips to Singapore or other regional hubs. Second, capacity management becomes simpler. If one Indian region experiences a surge in demand, Bedrock automatically shifts traffic to the other region without customer intervention. This automatic failover is a key selling point for production workloads that cannot tolerate downtime. However, the announcement leaves some questions unanswered. AWS has not disclosed pricing differences for India Geo inference compared to standard cross-Region profiles. It also has not clarified whether the India Geo profiles will extend to other OpenAI models or to other model providers on Bedrock. These gaps matter for enterprises building multi-year cost models. ## OpenAI's Multi-Cloud Strategy Takes Another Step OpenAI's presence on Bedrock is part of a deliberate multi-cloud expansion. The company already offers models through **Microsoft Azure** and its own API platform. By appearing on AWS, OpenAI reaches enterprises that have standardized on AWS infrastructure but want frontier model capabilities without switching clouds. This strategy reduces dependence on any single hyperscaler and broadens OpenAI's revenue base. For AWS, hosting OpenAI models is a defensive and offensive move. It defends against customers leaving for Azure to access OpenAI. It also attacks by making Bedrock a one-stop shop for nearly every major model family. The India Geo addition signals that AWS sees regulatory compliance as a competitive weapon, not just a feature. Enterprises evaluating this option should review the [Cross-Region inference section in the Amazon Bedrock User Guide](https://docs.aws.amazon.com/bedrock/latest/userguide/cross-region-inference.html) for technical details. Model cards for **GPT-5.6 Terra** and **Luna** provide performance benchmarks and context window specifications. Early adopters in Indian financial services and healthcare are expected to test these profiles in staging environments before year-end.