Skip to content
MukhyamantriMukhyamantri
TechnologyTamil Nadu

AI requests can now be processed in India, OpenAI opens local inference

OpenAI will let eligible Indian businesses have their API requests processed and model computations run inside the country, starting with the GPT-6 Luna model. The company says financial services firms with data localisation needs are the first target.

· · 1 min read

Chennai: OpenAI will now let businesses in India keep the whole AI loop at home, announcing local processing and inference for its API on Thursday.

Until now, eligible customers could store data in India, a facility the company introduced in May 2025. The new offering extends that to where the model processes information and generates its response, which is the sticking point for companies handling sensitive records.

Initially available for the GPT-6 Luna model, the service is aimed at organisations that have held back on deploying AI because of requirements around where their information is processed. Financial services customers with specific data localisation rules are the clearest fit, OpenAI said.

Under the arrangement, both the processing of an API request and the model computation that generates the response will take place in India for supported workloads, and supported customer content will be stored in the country. Customers get access by creating a new API project, selecting India for both storage and processing, and using the India regional endpoint.

The launch covers direct API usage and currently supports GPT-6 Luna input and output through the Responses and Chat Completions APIs. Other models and surfaces are expected to follow.

India is one of the fastest-growing markets for AI adoption, and Indian enterprises have been cautious buyers. Many run pilots and then stall at the purchase order stage, wary of compliance and of moving regulated data across borders.

Local inference also matters for latency. Round trips to overseas servers add delay, which hurts products such as voice assistants and customer support tools where a pause of a second is noticeable.

The move comes as domestic data centre capacity expands and as the government's IndiaAI Mission pushes shared compute for public institutions. Competing model providers have been building India capacity along similar lines.

For customers, the processing location is only one part of the compliance picture. Data-use terms, retention periods and the contractual promise on where records sit still need reading, technology advisers caution.

OpenAI has not said when the facility will cover other models in its range.