--- title: "OpenAI India Inference Residency for GPT-6 Luna, Explained" description: "API inference residency lets eligible businesses process GPT-6 Luna requests on servers in India, aimed at banks and other regulated buyers. The..." url: "https://bharat.digital/originals/openai-india-api-inference-residency-gpt-6-luna-data-localisation-analysis-nov-2026/index.md" canonical: "https://bharat.digital/originals/openai-india-api-inference-residency-gpt-6-luna-data-localisation-analysis-nov-2026/" published: "2026-11-13T01:05:00+04:00" updated: "2026-11-13T01:05:00+04:00" section: "Originals" type: "analysis" author: "bharat.digital desk" image: "https://bharat.digital/img/originals/openai-india-api-inference-residency-gpt-6-luna-data-localisation-analysis-nov-2026-og.jpg" language: "en-IN" site: "bharat.digital" --- # OpenAI now runs its AI inside India for some customers. Here is what that does, and does not, solve *API inference residency lets eligible businesses process GPT-6 Luna requests on servers in India, aimed at banks and other regulated buyers. The documentation shows strict eligibility rules, a price uplift and data that can still leave the country.* By bharat.digital desk. Analysis, AI & Tech. Published 2026-11-13T01:05:00+04:00. 5 min read. ![Abstract navy and indigo scene with a server rack and a neural network inside a translucent protective dome, with saffron and green data streams flowing into it](https://bharat.digital/img/originals/openai-india-api-inference-residency-gpt-6-luna-data-localisation-analysis-nov-2026-hero.webp) ## The question every bank asks first Ask an Indian bank, insurer or government department why it is still only experimenting with a frontier AI model, and the answer is often not about cost or accuracy. It is about geography. Where does a customer's loan file go when an AI model reads it? If the answer is a data centre in another country, many regulated teams stop there, however good the model is. On 12 November 2026 OpenAI moved to remove that objection for part of its business. It introduced what it calls API inference residency in India, which means that for supported workloads, both the handling of an API request and the computing that produces the answer now happen on servers inside the country, The Economic Times, Moneycontrol and Mint reported. Supported customer content is also stored in India. "Some Indian businesses have not fully deployed AI for important work because they need the information involved to be processed within the country," Nitin Bawankule, OpenAI's head of enterprise for India, told The Economic Times. He said local inference gives businesses more control over where supported workloads run and gives Indian developers and startups more room to build for their own and enterprise needs. ## Storage was the easy part OpenAI already let Indian customers keep some data at rest in India. The new step is about processing. Storage answers where a file sits when nobody is using it. Inference is the moment the model actually reads the prompt and writes the reply, and that is the moment a compliance team worries about most. OpenAI's own platform documentation is blunt about the difference: support for regional storage does not imply support for regional processing. Until now, Moneycontrol notes, OpenAI's residency options in India covered where customer data was stored. With this launch, India joins Australia and Japan as the latest markets where OpenAI runs the model itself in the region. The company says the feature is aimed at text tasks such as summarising documents, drafting replies and analysing information for employees, and at sectors such as financial services, where rules on where data is handled decide whether an AI application can be deployed at scale. ## The fine print that decides who can use it The launch is narrower than the headline suggests, and the details matter for anyone planning a project around it. At the start it covers one model, GPT-6 Luna, and only through two interfaces, the Responses API and the Chat Completions API. It applies to direct API usage, not to every OpenAI product. Customers have to create a new API project, choose India as the region for both storage and processing, and send requests to the India regional endpoint, which OpenAI's documentation lists as in.api.openai.com. The documentation adds conditions that the launch coverage mostly left out. Regional processing in Australia, India and Japan is available only to organisations with active Zero Data Retention, an OpenAI control that keeps customer content out of its abuse monitoring logs and has to be approved by the company. A lighter control called Modified Abuse Monitoring is not enough. Using data residency in any region other than the United States also requires approval for these controls and signing a Modified Retention amendment. In these three regions only Standard pay as you go usage is supported, and OpenAI's hosted tools are not available for regional processing. There is a price too. OpenAI's documentation says data residency endpoints carry a 10% uplift for eligible models released on or after 5 March 2026. And residency has limits on what it covers. The documentation says it does not apply to system data such as account details, usage statistics, billing and support requests, which may be processed and stored outside the selected region, or to data that leaves the region because of where the customer's own users or infrastructure are. ## A crowded field for local AI OpenAI is not the first global AI company to answer India's localisation question. In August, Anthropic announced plans to bring in country inference for its Claude models to India through Amazon Bedrock, AWS's managed service for building generative AI applications, Inc42 reported. OpenAI itself has also made a similar residency capability available for selected models on AWS Bedrock, according to Moneycontrol. That puts pressure on Indian model makers. Inc42 points out that for startups such as Sarvam AI, locally hosted inference has been an important way to stand apart from global players, and that the arrival of OpenAI and Anthropic in this space could narrow that advantage. The Indian companies still have other selling points, such as models built for Indian languages, but the simple pitch of keeping data in India is no longer theirs alone. ## Why OpenAI wants this market For OpenAI, the move fits a larger push. India is its largest market for ChatGPT, where more than 10 crore people use it every week, Inc42 reports, and the company is now looking to earn more from that base. Earlier this year it launched an initiative called OpenAI for India, covering AI infrastructure, enterprise adoption, skills and local partnerships, which included a partnership with Tata Group to build domestic AI data centre capacity, The Economic Times and Moneycontrol report. Paid enterprise API work, especially from banks and other large regulated buyers, is a far bigger prize than consumer chat. ## What to watch Three things will show whether this changes adoption or stays a checkbox. First, whether more models and features, including the hosted tools that are excluded today, are added to the India region. Second, how many Indian enterprises can actually clear the Zero Data Retention and contract requirements, since those approvals go through OpenAI's sales team. Third, how compliance teams at banks and government buyers read the remaining gaps, such as system data that can still leave the country. Local inference removes the biggest objection. It does not remove all of them, and the organisations most likely to care are the ones that will read the fine print first. ## Sources 1. [The Economic Times: OpenAI brings local inference to India, starting with GPT-6 Luna](https://economictimes.indiatimes.com/ai/ai-insights/openai-brings-local-inference-to-india-starting-with-gpt-6-luna/articleshow/135760461.cms) 2. [Moneycontrol: OpenAI API inference residency launches in India with local processing, in-country storage for GPT-6 Luna](https://www.moneycontrol.com/technology/openai-api-inference-residency-launches-in-india-with-local-processing-in-country-storage-for-gpt-6-luna-article-14078177.html) 3. [Mint: OpenAI goes local in India, enables in-country GPT-6 Luna API inference](https://www.livemint.com/ai/openai-goes-local-in-india-enables-in-country-gpt-6-luna-api-inference-how-will-it-benefit-businesses-11794476048553.html) 4. [Inc42: OpenAI Brings API Inference Residency To India To Deepen Enterprise AI Adoption](https://inc42.com/buzz/openai-brings-api-inference-residency-to-india-to-deepen-enterprise-ai-adoption/) 5. [CRN Asia: OpenAI brings API inference residency to India, addresses enterprise data localisation needs](https://www.crnasia.com/india/news/2026/openai-brings-api-inference-residency-to-india-addresses-enterprise-data-localisation-needs) 6. [OpenAI: Data controls in the OpenAI platform (documentation)](https://platform.openai.com/docs/guides/your-data) Written by the bharat.digital desk from the sources above. Figures are as reported by those sources.