Output involving tool calls cannot be rendered.

[wugxxx]

I have created an agent that can write a complete article based on keywords. However, when it comes to tool calls, there is no output on the page. Normal question-and-answer interactions can output normally.

I am using litellm to locally deploy LLM, and by checking the litellm log, the requests involving tool calls have complete output. What could be the possible causes for this?

[arkml]

Which model are you using?

[wugxxx]

@arkml I'm using Qwen3 - 32B deployed via litellm, and I'm encountering the same issue as with gpt-4o(proxy).

[wugxxx]

@arkml Okay, after I switched the model to Qwen2.5 - 32B, everything functioned normally. I guess it's because the thinking mode was enabled in Qwen3 - 32B.