OpenAI has introduced local processing and inference for its API in India, allowing eligible businesses and developers to run supported artificial intelligence workloads within the country. The move extends the company’s data residency capabilities beyond data storage to include the processing of API requests and the model computation involved in generating responses.
The development could help Indian enterprises, particularly those in regulated sectors such as financial services, assess the use of generative AI in business operations where the location of data processing is an important consideration.
The offering initially supports OpenAI’s GPT-6 Luna model through the Responses and Chat Completions APIs. For supported workloads, both the API request processing and the model inference take place in India. OpenAI also says supported customer content is stored in the country.
“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 – India, OpenAI. He added that local inference gives organisations greater control over where supported workloads are processed and could help address a barrier to AI adoption.
Moving beyond data storage
Data residency and inference residency address two different aspects of data handling. Data residency determines where information is stored, while inference residency determines where the computing required to process an AI request and generate a response takes place.
For enterprises, the distinction becomes important when applications handle customer records, financial information or other sensitive business data. Even if information is stored in a domestic data centre, the processing involved in generating an AI response could take place elsewhere, depending on the service architecture.
By offering in-country inference for supported API workloads, OpenAI is seeking to address this concern. The capability could support applications such as document summarisation, drafting responses and analysing information, where employees and business systems use AI to work with text-based content.
However, the residency commitment should not be interpreted as a guarantee that every part of an organisation’s AI workflow or all associated data processing remains within India. The regional processing capability applies to the supported models, endpoints and customer content covered by OpenAI’s terms.
Limited model and API support at launch
The India launch currently covers direct usage of the OpenAI API. Customers can create a new API project, select India for storage and processing, and use the corresponding regional endpoint.
At launch, GPT-6 Luna is the supported model, with access through the Responses and Chat Completions APIs. This means the capability is not a blanket regional processing option for every model, API service or tool available through the OpenAI platform.
OpenAI’s API documentation also sets out eligibility and service limitations for regional processing in India, Australia and Japan. These include a requirement for eligible customers to have Zero Data Retention enabled and the use of Standard pay-as-you-go processing. Hosted tools are not currently available for regional processing in these three regions.
These restrictions are relevant for enterprises evaluating the service, as applications that depend on additional API tools or processing features may need to be designed around the capabilities currently supported.
Part of OpenAI’s India expansion
The announcement builds on OpenAI’s broader India initiative, OpenAI for India, which covers AI infrastructure, enterprise adoption, skills development and local partnerships.
Earlier, the company announced a partnership with Tata Group to develop domestic AI data centre capacity. It has also made certain OpenAI models available with in-country inference through Amazon Web Services’ Bedrock platform.
The latest API launch gives developers and businesses using OpenAI directly another route to access regional processing. With data localisation becoming an important consideration in enterprise technology decisions, the availability of in-country inference could influence how organisations evaluate and deploy AI applications.
OpenAI said API inference residency is also available in Australia and Japan as part of the latest regional expansion.