OpenAI Data Residency: Power-User Guide to Regional API Projects

Last updated: November 19, 2026

OpenAI data residency lets an API project keep customer content stored, and in some regions processed, in a chosen region. Since November 18, 2026, eligible organization owners can turn it on themselves when creating a project, either as a fixed region or as per-request residency on a Global project, with a 10% price uplift on newer models.

Contents

What changed

OpenAI announced the change in its ChatGPT Enterprise and Edu release notes under the heading “Self-serve regional data residency for API projects,” dated November 18, 2026.

According to the note, eligible API organization owners can now choose regional data residency when they create a project. They can select an available fixed region or enable per-request data residency on a new Global project. The setup does not require a separate Zero Data Retention (ZDR) application, and the region picker shows whether each option covers storage only or storage and processing.

The Managing projects help article adds that only organization owners can create projects, and that eligibility for another API feature, including organization verification, does not by itself enable data residency.

Regions, storage and processing

The OpenAI data residency guide separates two promises. Storage means customer content is kept at rest in the region when an endpoint needs persistence, such as /v1/batches. Processing means inference also runs in that region, where the region supports it.

Region API hostname Storage Regional processing
United States us.api.openai.com Yes Yes
Europe (EEA + Switzerland) eu.api.openai.com Yes Yes
Australia au.api.openai.com Yes Yes (gpt-6-luna)
Japan jp.api.openai.com Yes Yes (gpt-6-luna)
India in.api.openai.com Yes Yes (gpt-6-luna)
Canada ca.api.openai.com Yes No
Singapore sg.api.openai.com Yes No
South Korea kr.api.openai.com Yes No
United Kingdom gb.api.openai.com Yes No
United Arab Emirates ae.api.openai.com Yes Yes (contact sales for access)

If a region does not support regional processing, OpenAI says it may process and temporarily store customer content outside the region to deliver the service. Image support outside the US requires additional approval. For requests to the US, Europe, Australia, India and Japan hostnames, OpenAI uses Cloudflare Regional Services so TLS termination happens inside the processing region.

How to set up a regional project

For a fixed-region project:

  1. Sign in to the API Platform as an organization owner of the intended organization.
  2. Open the project selector or the Projects page and select Create project.
  3. Enter a project name. Under Data and Inference Residency, select an available region and check whether it covers storage only or storage and processing.
  4. Review the pricing notice and the region, then select Create.
  5. Create an API key in the new project and send requests to that region’s hostname, such as https://eu.api.openai.com/v1.
  6. Confirm the project’s geography in its project settings.

The region is fixed at creation. Moving means a new project with separate resources.

Per-request residency on a Global project

If you want one project to serve several regions, follow the same steps but leave Data and Inference Residency on Global and select Enable per-request data residency before creating the project. The project stays Global.

Each request then picks its region by hostname. OpenAI’s Python example keeps one client and overrides the base URL per call:

client.with_options(base_url="https://eu.api.openai.com/v1").responses.create(
    model="gpt-6.1-terra", input="Reply with OK.")

The Go and Ruby SDK examples use a data residency option and send EU requests with a key from an EU-region project. Existing Global projects that already support per-request residency can keep their keys. A Global project without the option needs a new project.

Models, modes and pricing

Support differs by model, endpoint and mode:

  • GPT-6.2 Astra and GPT-6.2 Sol support US and EU data residency with Standard, Fast and Batch processing. Ultrafast works with US data residency only.
  • GPT-6.1 Sol, GPT-6.1 Terra, GPT-6 Sol and GPT-6 Luna support EU data residency with Standard, Flex and Batch. Fast mode supports EU residency for GPT-6.1 Sol, GPT-6.1 Terra and GPT-6 Luna, but not for GPT-6 Astra or GPT-6 Sol.
  • Australia, India and Japan support regional processing for gpt-6-luna on Standard pay-as-you-go only, through the Responses and Chat Completions APIs with text and image input and text output. Hosted tools are not available for regional processing there.
  • Tools such as web search, file search, Code Interpreter and the remote MCP tool list regional processing for the US and Europe only.

On price, the OpenAI API pricing page states that regional processing (data residency) endpoints are charged a 10% uplift for models released on or after March 5, 2026. For base rates, see our OpenAI API pricing guide and the GPT-6.2 Astra and Sol guide. For discounted completion windows, see our OpenAI background mode guide.

Fixed region vs. per-request residency

Question Fixed regional project Global project with per-request residency
Where requests go That region’s hostname Default hostname, or a regional hostname per request
Change region later No; create a new project Choose per request among supported regions
Best fit One workload that must stay in one region An app that serves users in several regions
Main check Storage only or storage and processing Endpoint and model support regional processing

For a dedicated alternative, our Cohere Model Vault guide covers dedicated GPU deployments with region choices. The GPT-6.1 Terra guide covers a model limited to US and EU residency.

Limitations and tips

  • System data is excluded. Account data, usage statistics, billing information, support requests and structured output schemas may be stored and processed outside the region.
  • Your own infrastructure counts. Residency does not cover data sent outside the region because of where your users or servers are, or data handled by third parties, including remote MCP servers.
  • Endpoint rules vary. Chat Completions cannot set store=true outside the US. Tracing for /v1/realtime is not yet EU residency compliant. Extended prompt caching in regions without regional processing may move content outside the region.
  • If the option is missing, confirm your organization and owner role, then contact OpenAI Support with your organization ID, project ID, chosen option, region and a sanitized error. Never share API keys.

OpenAI’s data controls guide explains abuse monitoring logs and the Zero Data Retention and Modified Abuse Monitoring controls, which still need prior approval.

Frequently asked questions

What changed in OpenAI API data residency on November 18, 2026?

OpenAI’s release notes say eligible API organization owners can now choose regional data residency themselves when creating a project. They can pick a fixed region or enable per-request data residency on a new Global project, and setup does not require a separate Zero Data Retention application.

Which regions does OpenAI offer for API data residency?

The data residency guide lists the United States, Europe (EEA and Switzerland), Australia, Canada, Japan, India, Singapore, South Korea, the United Kingdom and the United Arab Emirates. All offer storage. Regional processing is listed for the US, Europe, Australia, Japan, India and the UAE; UAE access requires contacting sales.

Does OpenAI data residency cost more?

Yes, for newer models. OpenAI’s pricing page says regional processing (data residency) endpoints are charged a 10% uplift for models released on or after March 5, 2026. Review the pricing notice in the project dialog before you create the project.

Can I change an OpenAI project’s region later?

No. OpenAI says a project’s residency region can’t be changed after creation. To use a different region, create a new project. Its resources are separate from the old project’s, and OpenAI warns against archiving a project to change its region, because archiving is permanent.

What is per-request data residency?

It is an option on a new Global project. You keep one Global project, then send each request that needs regional processing to a regional hostname such as us.api.openai.com or eu.api.openai.com with that project’s key. The endpoint and model must support regional processing.

Sources

Image: OpenAI logo, public domain, via Wikimedia Commons. OpenAI is a trademark of OpenAI.

Related answers

Leave a Comment