LLM Request Timeout

[rogerwelo]

Question

Hi,

Every now and then I encounter the issue that the LLM request gets stuck and thus the agent gets stuck without throwing an error.
Is there a way to set a timeout for LLM requests? And even better, the option to retry if one of them times out?

Additional Context

Pydantic AI 0.0.43
Python 3.11.7

[ItzAmirreza]

I'm also experiencing the same issue... For more context, I'm serving Qwen2.5-VL-7B using vLLM, tools do get called, but then response is always empty, and it just keeps doing the tool again and again, until it reaches timeout.

[Kludex]

How can I reproduce this? Please provide code and what I need to run.