gpt-4.1-mini Standard (regional) deployment in Australia East fails - model version deprecating, no newer version offered

Malik Waqar 20 Reputation points
2026-07-15T00:58:10.7066667+00:00

I need to deploy an Azure OpenAI chat model on the Standard (regional) deployment type in Australia East, because my workload has a data residency requirement to process data onshore within Australia. Global Standard and Data Zone deployment types are not suitable, since those can process data outside the region.

When I check Standard (regional) quota in Australia East for my subscription, the only chat-capable model that shows quota is gpt-4.1-mini (200K TPM). Every other current chat model (gpt-4o, gpt-4.1, gpt-5-mini, etc.) only appears under Global Standard or Global Batch, not Standard (regional).

When I try to deploy gpt-4.1-mini on Standard, it fails with:

"ServiceModelDeprecating: The model 'Format:OpenAI,Name:gpt-4.1-mini,Version:2025-04-14' is in deprecating state and cannot be used for new deployments."

The deployment dialog only offers this one version (2025-04-14). There is no newer gpt-4.1-mini version available to select.

My questions:

  1. Is there a newer, non-deprecating version of gpt-4.1-mini that can be deployed on Standard (regional) in Australia East? If so, how do I make it available in the version dropdown?
  2. If gpt-4.1-mini is fully deprecating, which current chat-completion model can be deployed on Standard (regional) in Australia East today for onshore processing?
  3. If no current chat model is presently available on Standard (regional) in Australia East, is there a known timeframe for one becoming available?

My subscription is pay-as-you-go. Usage is low (roughly 50 to 100 requests per month). Region and residency (Australia East, onshore processing) are firm requirements, so moving to a global deployment type is a last resort I am trying to avoid.

Any guidance on the correct current model + version + Standard deployment combination for Australia East would be much appreciated. Thank you.

Azure OpenAI in Foundry Models
0 comments No comments

Answer accepted by question author
Jerald Felix 18,760 Reputation points Volunteer Moderator
2026-07-15T01:59:49.4433333+00:00

Hello Malik Waqar,

Greetings! Thanks for raising this question in the Q&A forum.

The error you are hitting is not caused by anything wrong in your deployment configuration. It is a currently active platform-wide issue that many customers are reporting right now, and it results from two separate things overlapping in your specific scenario, which is worth separating clearly.

First, gpt-4.1-mini version 2025-04-14 is your only version because it is genuinely the only version Microsoft has ever shipped for that model. There is no newer version hiding in the catalog for you to select. Second, that version is officially listed as GA with a published retirement date of October 14, 2026, but many customers, across UK South, US regions, and Canada, are currently seeing the same ServiceModelDeprecating error on that exact version months ahead of its documented retirement date. This means Microsoft has moved it into the deprecating lifecycle stage early. Deprecating means no new deployments are accepted, while existing deployments keep working until the actual retirement date. This is why your dialog only shows one version and rejects it outright.

Here is how to work through it.

  1. Confirm this is deprecating, not fully retired

Query the Models API directly for your subscription and region to see the live lifecycle status rather than relying on the portal dialog alone:

GET https://management.azure.com/subscriptions/<subscription-id>/providers/Microsoft.CognitiveServices/locations/australiaeast/models?api-version=2024-10-01

If gpt-4.1-mini shows lifecycleStatus: Deprecated while the retirement date field still shows October 2026, that confirms the early deprecation behavior rather than a genuine end of availability.

  1. Check whether gpt-5-mini or gpt-5-nano are enabled for Standard (regional) in Australia East yet

These are the documented cost-tier replacements for gpt-4.1-mini, but their regional rollout has been uneven. At launch they were only available in East US 2 and Sweden Central for Global Standard and Data Zone, and other regions such as Canada Central and Canada East still do not have them under Standard (regional) as of now, with no published rollout date. Australia East may be in the same position. Confirm directly in your subscription:

az cognitiveservices account list-models \
  --name <your-foundry-resource-name> \
  --resource-group <your-resource-group> \
  --query "[?contains(model.name, 'gpt-5')]"

You can also check this in the Foundry portal under Operate then Quota, with the Show all toggle turned on, so you see every model available to your subscription in Australia East, including ones you have not deployed yet, not just ones with existing quota.

  1. If gpt-5-mini or gpt-5-nano are not yet listed for Standard (regional) in Australia East, do not switch deployment type as a workaround

Your earlier reasoning about Global Standard and Data Zone was correct and still holds. Data Zone deployment types are scoped to the US and EU data zones only, they do not exist for Australia, so neither option satisfies onshore residency. Provisioned (PTU) is technically a separate deployment type from Standard and sometimes gets earlier regional access to newer models, but at your volume, roughly 50 to 100 requests a month, the PTU minimum commitment will not be cost effective, so it is worth checking only if you are blocked long term.

  1. Open an Azure Support ticket rather than continuing to retry deployment

Since this is a platform-side lifecycle and rollout issue rather than something fixable from your side, file a ticket under Azure Portal then Help + support then Create a support request, scoped to the Azure OpenAI or Foundry resource in question. In the description include:

  • The exact error text and the affected model and version: Format:OpenAI,Name:gpt-4.1-mini,Version:2025-04-14
  • A note that the published retirement date for this version is October 14, 2026, but new Standard deployments are already being blocked
  • Your Subscription ID and the resource ID, for example /subscriptions/<subscription-id>/resourceGroups/<rg>/providers/Microsoft.CognitiveServices/accounts/<resource-name>
  • A request to confirm whether gpt-5-mini or gpt-5-nano Standard (regional) is planned for Australia East, and if so, when
  1. Track ongoing announcements

Since Microsoft does not publish forward-looking rollout dates for adding models to specific Standard regions, monitor azure.microsoft.com/updates filtered by Azure OpenAI, and enable Azure Service Health alerts for Health advisories on your subscription so you are notified automatically when gpt-5-mini or gpt-5-nano land in Australia East for Standard.

If this answer helps you kindly accept the answer which will help others who have similar questions.

Best Regards,

Jerald Felix

Was this answer helpful?

1 person found this answer helpful.
0 comments No comments

1 additional answer

Sort by: Newest
  1. Malik Waqar 20 Reputation points
    2026-07-15T06:10:56.46+00:00

    Hi @Jerald Felix

    Thank you, this is really helpful and it confirmed what I was seeing. You are right that gpt-4.1-mini 2025-04-14 is the only version, and that it has moved into the deprecating stage early so new Standard deployments are blocked. I also checked gpt-5-mini and gpt-5-nano, and neither is available on Standard (regional) in Australia East yet either, so there is currently no current chat model I can deploy on the onshore Standard tier in that region.

    Based on your answer I have set up Azure Service Health alerts so I will be notified automatically when a suitable model becomes available on Standard in Australia East, and I will switch to it then. In the meantime we have decided to proceed on Global Standard, with the residency trade-off understood.

    Marking your answer as accepted. Thanks again for the clear explanation and the API and CLI checks, they made the situation much easier to confirm.

    Best regards,

    Malik Waqar

    Was this answer helpful?

    0 comments No comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.