[Issue] Tool Calling Failures on Groq LLMs via Pydantic-AI (OpenAI baseline vs Groq)

Forum
2

As a follow-up to my earlier post, I wanted to share one more trace and a reproducible notebook.

Here’s another example from Logfire:

To make this easier to verify, I’ve also prepared a notebook that reproduces the runs I described.

Curious if anyone else has tested this model specifically, or if there are known workarounds to improve tool call reliability on Groq.

2 Likes