The distinction sounds technical and is not. If your agency, health service or regulated business is adopting the OpenAI API and someone has written "data stays in Australia" in the compliance documentation, that sentence is half true, and the half that is false is the half your privacy officer probably cares about.
What OpenAI's documentation actually says
OpenAI's data controls documentation distinguishes two things a region can offer. Regional storage keeps customer content at rest inside the region. Regional processing runs the model itself on in-region infrastructure. Only two listed regions get both: the United States and Europe. Australia, alongside Canada, Japan, India, Singapore, South Korea and the UK, is listed "Storage: Yes, Processing: No". The document then says the operative sentence plainly: "If your selected Region does not support regional processing... OpenAI may also process and temporarily store Customer Content outside of the Region to deliver the services." Using the Australian endpoint also requires approval for abuse monitoring controls and a modified retention amendment, OpenAI's conditions, listed on the same page.
| Region | Data stored in region | Prompts processed in region |
|---|---|---|
| United States | Yes | Yes |
| Europe (EEA + Switzerland) | Yes | Yes |
| Australia (au.api.openai.com) | Yes | No |
| Canada, Japan, India, Singapore, South Korea, UK | Yes | No |
Source: OpenAI, "Your data" API documentation, region table, accessed 11 July 2026. The table above simplifies OpenAI's per-service matrix; some individual endpoints have further limitations listed in the source.
And what it costs
OpenAI's pricing page states: "Regional processing (data residency) endpoints are charged a 10% uplift for models released on or after March 5, 2026, that are eligible for data residency." That covers the current GPT-5.6 family, priced (standard tier, USD per million tokens) at $5 input / $30 output for Sol, $2.50 / $15 for Terra and $1 / $6 for Luna. In other words, for an Australian buyer the residency endpoint costs 10 per cent more and still does not buy onshore processing, because no onshore processing exists to buy. All prices are in US dollars; your AUD cost floats with the exchange rate on top.
What this does and does not mean for compliance
None of this makes the Australian region pointless. Storage residency is a real control: content at rest, files, batch inputs and stored completions can stay onshore, which answers some records-management and breach-exposure questions. What it does not answer is the sovereignty question people usually mean: while it is being processed, your prompt, and whatever personal information rides in it, transits infrastructure in another jurisdiction. Under the Privacy Act's APP 8 that is a cross-border disclosure question, and it does not disappear because a region selector says Australia.
Our advice, offered as considered opinion on the documents above: write the two-line version into your assessment, "data at rest in Australia; processing offshore per OpenAI's documentation", cost the 10 per cent uplift against what storage residency is actually worth to you, and treat any vendor or integrator who tells you the Australia region means Australian processing as having not read the page. It is one page. It is linked below.