Amazon Brings OpenAI's Latest AI to India — Data Stays Put

Amazon is now letting businesses in India use OpenAI's newest AI models, called GPT-5.6 Terra and Luna, through Amazon Bedrock — its service for building AI-powered apps. The big twist: all data processing happens inside India, not in other countries. That's a huge deal for banks, hospitals, and government agencies that are legally required to keep sensitive information within India's borders. Companies can now use world-class AI to analyze financial reports, scan medical records, or review legal contracts without worrying about data residency violations.

These models are powerful. They can read a million tokens at once — think of tokens as chunks of words. A million tokens is roughly the length of several long books. They also handle both text and images, so you could upload a contract with scanned signatures or a medical report with x-ray images, and the AI will process everything in one go. Amazon says the processing routes between two regions in India, Mumbai and Hyderabad, and never leaves the country. Data is encrypted during travel, and Amazon doesn't store inputs or outputs by default, which adds another layer of safety.

For businesses, this removes a painful tradeoff. Before, if you wanted the best AI, you often had to send data to servers abroad. Now Indian companies can have both: top-tier intelligence and local data control. Amazon also offers a 'global' option that routes requests anywhere for maximum computing power, but advises Indian businesses with local requirements to use the India-specific setting. This launch makes advanced AI more accessible to regulated industries that had to sit on the sidelines.