News in short

• OpenAI has introduced API inference residency in India for eligible businesses and developers.
• The service processes supported AI requests and model computation within India.
• GPT-6 Luna is currently the only model supported for regional processing in India.
• Organisations need active Zero Data Retention (ZDR) and an India-based API project.
• The rollout also covers Australia and Japan.

OpenAI has expanded API inference residency to India, giving eligible businesses more control over where their AI workloads run. The new option allows supported requests and the model computation behind their responses to take place within the country.

The move could help Indian organisations that face strict data localisation requirements. It is particularly relevant to sectors such as financial services, where companies often need tighter control over customer information.

OpenAI brings local API processing to India

OpenAI says eligible organisations can now choose India for both data storage and processing in a new API project. They must also use the corresponding India regional API endpoint. The arrangement covers supported customer content and the model computation required to generate responses.

However, the rollout has important limits. India-based regional processing currently supports GPT-6 Luna. It works through the Responses and Chat Completions APIs, with text and image inputs and text output. The regional setup supports Standard pay-as-you-go processing. Hosted tools are not available for these workloads.

Who can use OpenAI inference residency in India?

The option is not available to every API customer by default. OpenAI’s documentation says organisations in India, Australia and Japan need active Zero Data Retention (ZDR) to use regional processing. They must also create a project in the selected region and configure the matching regional endpoint.

This distinction matters because regional data storage does not automatically mean that model inference happens in the same location. OpenAI’s new option covers both for supported workloads, subject to its eligibility conditions.

What can businesses do with local AI processing?

Businesses can use supported API workloads for tasks such as summarising documents, drafting responses and analysing information. For example, a financial services company could explore internal AI assistants while keeping eligible processing within India.

OpenAI says the move gives Indian developers and startups more room to build products for enterprise needs. Nitin Bawankule, Head of Enterprise for India at OpenAI, said local inference gives organisations greater control over where supported workloads run.

The expansion also reaches Australia and Japan. It adds three Asia-Pacific markets to OpenAI’s regional processing options, although the available models and features vary by region.

Why the India launch matters

Data localisation can influence whether a company moves an AI project from testing into production. By offering in-country inference, OpenAI is addressing one practical barrier for organisations with stricter compliance needs.

Still, businesses should review the ZDR requirement, supported APIs and tool limitations before migrating workloads. The India rollout is a targeted option for eligible customers, not a blanket promise that every OpenAI service or model now processes data locally.

For more on OpenAI’s enterprise push, read our report on Businesses in ChatGPT and its AI agents. Developers can also check OpenAI’s official API data controls documentation for regional eligibility and setup details.