OpenAI has launched local processing and inference for its API in India, allowing eligible businesses and developers to process supported AI requests within the country. The service is available initially for the GPT-6 Luna model, and it extends the company's existing India data residency offer from where customer content is stored to where it is actually processed.

OpenAI announced the launch on Thursday 12 November. Until now, its India residency option covered customer content stored at rest in the country. For supported workloads, both the processing of an API request and the model computation that generates its response will now take place in India, with supported customer content also stored in the country.

Why location matters to Indian businesses

For some Indian organisations, where AI processes information can decide whether an application moves into production at all. OpenAI said the change is especially relevant to customers in financial services with specific localisation requirements, giving them the chance to assess how to deploy AI across their operations.

"Some Indian businesses have not fully deployed AI for important work because they need the information involved to be processed within the country," said Nitin Bawankule, head of enterprise for India at OpenAI. "Local inference gives them more control over where supported workloads run and helps address a practical barrier to adoption. It gives more businesses a way to put OpenAI's tools to work, and gives Indian developers and startups more scope to build for their own, and enterprise needs."

What is covered, and what is not

The launch covers direct API usage only. For now, local processing supports GPT-6 Luna input and output through the Responses and Chat Completions APIs. It does not change the experience of regular ChatGPT app users, whose requests are not covered by the new arrangement, and no timetable has been given for extending regional processing to other models.

Access is subject to eligibility conditions. OpenAI's developer documentation states that organisations using regional processing in India must have Zero Data Retention enabled, the setting under which the company does not retain API request and response data, and must use Standard pay-as-you-go processing. Hosted tools, the company-provided tools that can be invoked during an API call, are not currently available for regional processing in India. Customers can get started by creating a new API project with India selected for both storage and processing, and using the corresponding India regional endpoint.

Part of a wider India push

India becomes the third market in which OpenAI offers inference residency, joining Australia and Japan, according to reporting of the launch. The announcement forms part of OpenAI for India, an initiative introduced earlier this year spanning AI infrastructure, enterprise adoption, skills development and local partnerships, which included a partnership with Tata Group to develop domestic AI data centre capacity. OpenAI has also made a similar residency capability available on certain models in AWS Bedrock.

The launch is aimed at organisations handling sensitive information, including banks and other regulated businesses, that need to show where their data is processed as well as where it is stored. OpenAI has not disclosed an investment figure for the India processing capability, nor the number or locations of the data centres involved.

Sources: The Hindu BusinessLine and OpenAI — Data controls in the OpenAI platform.