API Rate Limit Exceeded Error When Interacting with LLM

[ANKIT7-DEL]

While using the LLM integration

LLM error: {
"message": "API rate limit exceeded",
"request_id": "af8869793862ad7ab720bb7f5438dd6a"
}
change introduces a retry mechanism that will delay the response when rate limits are hit. The default is 3 retries with exponential backoff starting at 1 second.

[ANKIT7-DEL]

[phuducnguyen]

Rate limits are a common part of integrating any external API. This error dialog is a solid improvement that helps the system behave more gracefully under load.

This is a normal situation when working with LLM APIs and not an issue with your implementation.

Nice catch addressing this early 👍

[dvanco]

So what is the expected user reaction here? Do I simply allow shai to process or do I need to wait some determined amount of time and request execution again? From the look of the error it's "done" and I'll need to tray again later.

I waited an hour and attempted the action again (building a simple folder structure for a website) and it failed again immediately simply trying a mkdir.

I don't see any other doc or README that details rate limits. Is there a way to avoid being rate-limited? (note I am on the US infra, not EU)

EDIT - US Support informs me that the infra has not been prepped for AI Endpoints as of yet (and have no ETA) - this would seem to be some cogent information for the docs both here and there.