AI Chat Template - Slows down quickly as the chat runs

[345Nathan-Harrison]

Looking for some help regarding the AI chat template app. The overall issue is when I run my application, the first question gets answered really quickly and then after that it really slows down in how responsive it is.

My application is a simple web based AI chat (using the new AI template available from Microsoft). The agent using AI Search to assist in understanding a few documents I have given it. All very simply in context.

In AI Foundry I've wired up gpt-4o and text-embedding-ada-002 models into the application which are required.

I have looked at the rate limits which seen fine (not entirely sure though).

I believe it has to do with something regarding using the previous answers/ messages to continue the conversion - but again not entirely sure. It would at least make sense as to why it slows down so much after the first question gets responded to.

Any ideas, or if you've come across this before, would be really appreciated - thank you!

[345Nathan-Harrison]

Switched from an MSDN subscription to a non-msdn subscription which appears to have fixed it and its quicker. Wondering why that is the case?

[stephentoub]

Switched from an MSDN subscription to a non-msdn subscription which appears to have fixed it and its quicker.

Do you have different rate limits set on the Azure accounts you're connecting to?

My guess is you were being rate limited in one case and aren't in the other.

[345Nathan-Harrison]

I'll have a look and examine each of these finer details and check - Thanks