OpenAI has added the UAE to its inference residency programme — one of only three regions worldwide, alongside the United States and Europe — where customers can keep AI model inference running inside the country.
Eligible API, ChatGPT Enterprise and ChatGPT Edu customers can now constrain GPU inference to UAE-based engines, so prompts and responses are processed in-region rather than shipped abroad.
The launch extends the UAE data residency option OpenAI announced last September, which kept customer content stored at rest in-country. Inference residency goes further: the GPUs that run the models, and the prompts and responses flowing through them, stay in the Emirates too. For enterprises, public-sector bodies and education providers working under local compliance rules, that is the difference between data sitting in Dubai and the work happening in Dubai.
What UAE inference residency means
UAE policy is pushing the same way: the National AI Strategy 2031, the new Federal Authority for Artificial Intelligence and Data, and data-localisation requirements spreading across government, healthcare and financial services. For organisations in those categories, in-country GPU execution is what data residency alone did not offer.
“Organizations across the UAE are adopting AI at a remarkable pace, supported by the Government’s clear long-term vision and commitment to technology leadership. With the launch of UAE Inference Residency, we are giving eligible customers greater control over where AI workloads are executed, building on our existing residency offerings and supporting institutions that need stronger regional assurances as they scale AI adoption.”
Farouk El Hamzawi, head of enterprise, OpenAI MENA
There are limits to the UAE option. Under UAE inference residency, GPT-5.2 is currently the only model available in ChatGPT and for API inference; image generation, internal search, ChatGPT Work and improved memory are unsupported, and GPT-Live is not available either. The newer GPT-5.6 family does not support UAE inference residency workloads, per OpenAI’s help centre. External web search still works, and non-GPU processing such as authentication and analytics can still leave the region.
The move lands as Stargate UAE — the 1GW compute hub built with G42, Oracle and SoftBank — starts coming online in phases, per Middle East AI News. OpenAI has been pushing deeper into the region’s enterprise market all year, from its enterprise deployment unit to UAE-targeted security offerings. The full eligibility and model details are on OpenAI’s help centre.
What is the difference between OpenAI data residency and inference residency?
Data residency keeps customer content stored at rest in a specific region. Inference residency goes further: model inference, the GPU execution that produces responses, also runs on hardware inside that region. Inference residency requires data residency to be enabled in the same region.
Which regions support OpenAI inference residency?
Three regions: the United States, Europe (EEA plus Switzerland) and the United Arab Emirates. OpenAI says it plans to expand to more regions over time.
How much does OpenAI inference residency cost?
Data residency is included at no additional cost on ChatGPT Enterprise and Education plans. Inference residency rides on the same eligibility; customers unsure of their status are directed to OpenAI’s sales team.
