I created a fresh project in Azure AI using the portal, with all new Azure resources associated with the project.

I created a deployment of gpt-4o-mini with the default deployment options.

Deployment type: Global Standard

Rate limit (Tokens per minute): 8,000

Rate limit (Requests per minute): 80

Provisioning state: Succeeded

Looking at the metrics of this model deployment, I see:

As you can see, I am far from any of the rate limits.

I create a new Agent in the portal (using Assistants API within Azure AI Agent Service preview).

I start a playground session for this agent in the portal, a thread is created and I enter (any) user input.

As you can see, I am getting an error message claiming I have hit a rate limit.

This makes no sense to me. I can start a playground session on the same model deployment itself and consuming that model there, there are no rate limiting issues.

What is going on here? I don't seem to be able to make any progress with AI Agent Service. What am I missing?

Thank you.