> ## Documentation Index
> Fetch the complete documentation index at: https://docs.make-pretty.com/llms.txt
> Use this file to discover all available pages before exploring further.

# AI provider overloaded

> What to do when the AI provider is temporarily overloaded and your request can be retried.

> **The AI provider is temporarily overloaded. Wait a moment, then try again.**

## What happened

The model provider could not process your request because its service was overloaded or
temporarily unavailable.

## What you can do

1. Wait a minute, then press **Retry**.
2. If the error keeps returning, choose another available model and send the request again.
3. If every model fails, [contact Pretty support](/help/support) with the time of the error and the
   message ID.

Provider overloads usually clear without a settings change. Repeated failures across several
minutes may point to a provider incident or a problem with your workspace's connected AI service.

## For admins

Check the provider's status page first. If your workspace uses its own provider or LiteLLM proxy,
also check the proxy logs and upstream provider response. Pretty shows this message for temporary
provider errors such as `server_error`, `server_is_overloaded`, and `vector_store_timeout`. Direct
provider connections also use it for HTTP `5xx` responses. A LiteLLM gateway response with HTTP
`502`, `503`, or `504` is treated as a proxy outage instead.

If the proxy itself cannot be reached, Pretty shows a different message. See
[AI service isn't available](/errors/ai-provider-unreachable).