Azure Open AI model update

Question

My current setup in Foundry portal Azure Open AI includes gpt-4o-mini

Version - 2024-07-18

Retirement Date - Oct 1, 2026 2:00 AM

Deployment Type - Global Standard

I have also this version upgrade policy: "Once a new default version is available"

What does it mean in case of model retirement? Will there be an auto update to new model and when? Once model will retire or earlier? Is there any exact date?

Answer 1

Hello Mateusz Chalaba,

Welcome to Microsoft Q&A and Thank you for reaching out.

I understand that you’ve deployed GPT-4o Mini** (version 2024-07-18)** with the “Once a new default version is available” upgrade policy. Here’s what that means and how it relates to the retirement date.

Update policy mapping

The Once a new default version is available policy corresponds to Azure’s Auto-update to default setting.

When Microsoft designates a newer version of the same model family as the default, your deployment will automatically switch to that version.

How the upgrade works

If a new default version of GPT-4o-mini becomes available before the retirement date, your deployment will automatically upgrade to that newer version.

This upgrade typically happens before the current version retires, once the new version is marked as the default.

Timing of the update

The auto-upgrade occurs when a new default version is published, not necessarily on the retirement date.

Microsoft usually provides advance notice (typically at least ~2 weeks) before a version becomes the new default.

Retirement behavior

• Your current version (2024-07-18) has a retirement date of Oct 1, 2026.
• With Auto-update to default, your deployment will likely move to the newer default version before Oct 1, 2026, once it becomes available.
• If you prefer to remain on the current version until it retires, you could instead use the “Upgrade when expired” policy. In that case, the upgrade would occur after the retirement date.

Exact upgrade timing

There isn’t a fixed public date for when the upgrade will occur. The automatic upgrade happens when:

A new version of GPT-4o-mini is released, and

That version is designated as the default version.

Once both conditions are met, deployments using the Once a new default version is available policy are automatically updated.

Please refer these

Model Retirements & Deprecations: https://learn.microsoft.com/azure/ai-services/openai/concepts/model-retirements#current-models

Working with Models & Update Policies: https://learn.microsoft.com/azure/ai-services/openai/how-to/working-with-models?tabs=powershell#model-updates

New Model Announcements (AI + ML Blog): https://azure.microsoft.com/blog/category/ai-machine-learning/

Even with automatic upgrades enabled, many teams test the newer model version in a separate deployment first before production workloads switch automatically. This helps confirm compatibility and performance before the upgrade occurs.

I Hope this helps. Do let me know if you have any further queries.

---

If this answers your query, please do click Accept Answer and Yes for was this answer helpful.

Thank you!