Unexpected region mismatch in Azure OpenAI Service.

Question

Hey everyone,

I accessed my Azure OpenAI Service resource via the Azure Portal and deployed the gpt-4o model using Azure AI Foundry. My Azure OpenAI Service was set to the japaneast region, but here's what happened:

• The model got deployed in a different region (eastus) than my Azure OpenAI Service.
• An Azure AI Services resource was automatically created in the same region as the deployed model. It looks like the model ended up being deployed on that AI Services resource.
  • The resource name follows this pattern: "{shortened username}-{8 alphanumeric digits}-{deployment region (eastus)}" which makes it seem like it was created mechanically.

As of October 2024, I remember that model deployments to Azure OpenAI Service were supposed to be limited to the same region. Has there been a change in the specs somewhere? Any insights would be appreciated!

Thanks!

Answer 1

Hello Isya,

The deployment behavior you’ve experienced is due to regional model availability and recent updates in Azure’s deployment specifications. Previously, Azure OpenAI model deployments were restricted to the same region as the Azure OpenAI Service resource. However, with Azure AI Foundry, model deployments are now determined by regional availability rather than being limited to the OpenAI Service’s region. If a model, such as GPT-4o, is not available in your selected region (japaneast), it will be deployed in a supported region (eastus in this case). Additionally, Azure AI Foundry automatically provisions an Azure AI Services resource in the same region as the model deployment, which explains the system-generated resource name pattern you observed.

Kindly refer this document Chat completions,

I hope you understand. And, if you have any further query do let us know.

---

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

Thank you!