OpenAI launches local AI processing for API users in India
Eligible customers can now run GPT-6 Luna requests on domestic servers, helping organizations meet strict data residency requirements.
The short version
- Eligible API users can now process GPT-6 Luna requests inside India for both computation and storage.
- The update expands on a 2025 residency expansion that previously only covered data storage, not processing.
- Financial services are a primary target for these features, specifically for tasks like document summarization.
- System data like billing and usage statistics remain excluded from these regional residency commitments.
- The service is available via a new project setup that requires contacting OpenAI sales for access.
Why it matters
This allows Indian companies and research institutions to use advanced AI models while keeping sensitive data within national borders. It addresses a major hurdle for industries with strict regulatory requirements regarding where their data is handled.
Domestic processing for India
OpenAI announced on November 12, 2026, that it is enabling local API inference for eligible customers in India. This means that supported GPT-6 Luna requests can now be processed and computed entirely within the country. The move is designed to ensure that both the model's computation and the storage of customer content happen locally.
This update builds upon a residency expansion that occurred in late 2025. While that previous move allowed for regional storage, the current announcement specifically adds the processing component. This distinction is important because it addresses the full lifecycle of a data request rather than just where the final result is saved.
The launch is part of the broader OpenAI for India program. This initiative covers various aspects of the company's presence in the region, including infrastructure, enterprise adoption, and local partnerships. By providing local inference, OpenAI aims to make its tools more accessible to organizations with specific geographic constraints.
Targeting high-security sectors
The new capability is particularly relevant for sectors like financial services. These industries often face strict regulations regarding how they handle internal information. The company suggests that these tools could be used for drafting responses, summarizing documents, and analyzing internal data.
Nitin Bawankule, OpenAI’s head of enterprise for India, noted that some businesses had avoided AI because of these location requirements. He stated that local inference provides organizations with greater control over where their workloads are handled. This is intended to help Indian developers and startups deploy AI more freely.
However, the announcement does not automatically apply to every OpenAI product or model. It specifically covers direct API use for supported workloads. Users must check the specific data-residency documentation to see which models and tiers are eligible for the India configuration.
Technical boundaries and controls
OpenAI’s residency guide clarifies that certain types of data are excluded from these regional commitments. System data, such as billing information, usage statistics, and account details, are not governed by these local processing rules. Additionally, transfers caused by the location of a customer's own infrastructure are also excluded.
The company distinguishes between abuse-monitoring logs and the application state used to support specific features. While API data is not used to train models by default, customers can opt in if they choose. Furthermore, features like Zero Data Retention require prior approval and are separate from the residency settings.
For research offices, these distinctions mean that data management plans must be precise. Teams need to record location, retention, and training permissions as separate variables. One single setting does not serve as evidence for all three of these different data governance requirements.
Implementation for institutions
Institutions can now evaluate a more specific service configuration for their AI projects. An application for local document analysis can now be assessed against a named model and a specific processing location. This provides a concrete framework for university and research teams to discuss with the provider.
The NIST AI Risk Management Framework suggests connecting these AI policies to existing data governance. This involves documenting requirements, defining model testing, and establishing incident-response processes. By using these guidelines, organizations can better map the new local inference option to their internal safety standards.
The contractual side of this deployment involves a Data Processing Addendum. This document helps clarify the responsibilities between the provider and the customer. It is important to distinguish between a data-sharing agreement and a data-processing agreement when reviewing these new domestic options.
Access and next steps
The launch is currently available to eligible API users who want to set up a new project. To activate these features, customers must select India for both storage and processing during the setup process. The official documentation directs interested customers to contact the sales team to gain access to these specific regional endpoints.
While this launch is a significant step for domestic data handling, it does not automatically guarantee compliance with all Indian laws. It provides the technical infrastructure for local processing, but the specific use case must still be approved by the institution. Organizations must still determine if the tool fits their unique research or business needs.
This development follows OpenAI's partnership with the Tata Group for domestic data-center capacity. It also sits alongside the availability of selected models through AWS Bedrock. Together, these initiatives form a larger strategy to support the growth of AI infrastructure and enterprise adoption within the Indian market.
What to watch
- Whether more Indian industries, beyond finance, begin large-scale deployments.
- If OpenAI expands local inference to other regions with similar data residency laws.
- How Indian research institutions integrate these local processing options into their data management plans.
Try it
Check OpenAI's data-residency documentation to see which specific GPT-6 Luna tiers are eligible for India processing.
Sources and links
- OfficialAPI data-controls documentation developers.openai.com
- OfficialOpenAI’s data-residency documentation developers.openai.com
- OfficialOpenAI’s November 25, 2025 residency expansion openai.com
- GovernmentNIST AI Risk Management Framework’s Govern playbook airc.nist.gov
- ReportingCASRAI: the article we sourced
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by CASRAI · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
