Unable to deploy and test resource

Question

Hello,

I am following the official Microsoft Learn exercise at the link below and have been unable to complete it after several hours of troubleshooting:

https://microsoftlearning.github.io/mslearn-ai-studio/Instructions/Exercises/01-Explore-ai-studio.html

Subscription ID: c570c2b7-cd98-403d-a6c7-eb31d74b8f44

Resource: AI-102-Training-TW-Test

Resource Group: Training

Region: East US

Root cause of issue:

When attempting to deploy gpt-4.1 or gpt-4o-mini through Microsoft AI Foundry, all available regions in the region selector are greyed out and unselectable. When I attempt to proceed without a region, I receive the following error:

'Failed to deploy model gpt-4o-mini — InvalidCapacity: The specified capacity 0 of account deployment should be at least 1 and no more than 1000000.'

Trace ID: d4bd1531-70fb-4dfd-bdd2-2fd6318c8892

Client request ID: 1891b062-b422-4b3e-939d-71e15c158491

Service request ID: 66cc84f9-e6dd-4945-9844-2b692375ef09

The tokens per minute rate limit slider is locked at 0 and cannot be adjusted, and the self-service quota request option is greyed out and unavailable. I recently upgraded from Azure for Students to a pay-as-you-go subscription specifically to resolve this issue, but the problem persists.

I am preparing for the Microsoft AI-102 Azure AI Engineer Associate certification exam and cannot complete any of the required hands-on lab exercises without this quota being granted. This has been a significantly negative experience. I am requesting a quota of at least 10,000 tokens per minute for gpt-4o-mini in East US so I can proceed with my certification training.

Thank you for your prompt assistance.

Answer 1

Timothy Wade hi,

I think it is not a deployment bug it means ur subscription still has zero quota for those models in that region. When the slider is stuck at 0 and regions are greyed out, it means there is no capacity assigned to ur subscription for gpt 4.1 or gpt 4o mini, so Foundry cannot deploy anything and defaults to capacity 0 which then throws the InvalidCapacity error.

Upgrading from Azure for Students to Pay As You Go does not automatically grant OpenAI model quota. OpenAI quota is granted separately and sometimes takes time after subscription changes.

What u need to do is request quota from the Azure OpenAI quota page, not from the generic capacity blade. Go to Azure Portal, search for Azure OpenAI, open the quota management page and submit a request specifically for gpt 4o mini in East US. If the self service request is greyed out, it usually means ur subscription is either too new, not fully verified, or requires manual approval. If that option stays disabled, open a support ticket under Azure OpenAI and request quota increase manually. Include the subscription ID, region, and requested TPM like 10000. There is no way to deploy the model until quota is granted.

So the root issue is zero assigned TPM for those models not a lab configuration problem. Once quota is approved regions will become selectable and the slider will unlock.

rgds,

Alex