OpenAI introduced in-country processing for GPT-6 Luna API workloads in India on 12 November, extending its existing local data residency option beyond storage, Analytics India Magazine reported. Eligible businesses and developers can use the capability to run supported requests and store supported customer content within India. For organisations whose AI applications require domestic processing, the change covers the computation that produces a response as well as the location of stored data.
Key points
- OpenAI’s India option covers supported GPT-6 Luna requests through its Responses and Chat Completions APIs.
- Regional processing requires an organisation to have active Zero Data Retention.
- The India region supports standard pay-as-you-go usage, but not other processing modes.
- India joins Australia and Japan among the latest locations for OpenAI’s API inference residency.
GPT-6 Luna requests stay in India
For supported workloads, OpenAI says the API request and the model computation used to generate its response take place in India. Supported customer content is also stored there. The distinction matters to businesses that need to keep processing within the country: choosing a domestic storage location alone did not provide the same option for inference, the computation that generates a model’s answer.
The initial scope is direct API use of GPT-6 Luna through Responses and Chat Completions. According to CRN Asia, the supported work includes summarising documents, drafting responses and analysing information. Organisations enable the option by creating a new API project, choosing India for both storage and processing, and sending requests to the India regional endpoint.
Nitin Bawankule, OpenAI’s Head of Enterprise – India, said some Indian businesses had held back important AI deployments because the information involved needed to be processed domestically, PTI reported. OpenAI identified financial services as a sector where processing location can affect whether an application enters production. The launch gives eligible customers a way to meet that location requirement for the supported API workloads.
Zero Data Retention limits access
Eligibility narrows the offering. CRN Asia reported that OpenAI’s API documentation makes India regional processing available only to organisations with active Zero Data Retention. That condition applies alongside the model and API restrictions, so an organisation’s ability to use OpenAI’s API does not by itself give it access to processing in India.
The commercial route is standard pay-as-you-go usage. Other processing modes are not supported in the India region, BestMediaInfo reported, citing OpenAI’s API documentation. That leaves customers seeking local processing to use the supported mode when configuring workloads around the regional endpoint.
CRN Asia reported that systems integrators and software developers could work with customers to connect locally processed requests to existing applications. It also reported that OpenAI had not announced a specific channel programme or commercial arrangements for partners with the launch. Any such work would therefore sit outside a newly announced partner programme.
OpenAI expands its India infrastructure options
The API launch follows OpenAI for India, an initiative introduced earlier in 2026 covering infrastructure, enterprise adoption, skills and partnerships. As part of that initiative, OpenAI announced a partnership with Tata Group to develop domestic AI data centre capacity, Analytics India Magazine reported. The local processing option addresses where eligible API requests run, while the Tata partnership concerns capacity planned for development.
OpenAI has also made in-country inference available for certain models through Amazon Web Services’ Bedrock platform, according to CRN Asia. That route is separate from the newly available direct OpenAI API option for GPT-6 Luna. India, Australia and Japan are among the latest locations added to OpenAI’s API inference residency offering.
