How are you running AnythingLLM?

All versions

What happened?

When using a reasoning model that returns no thoughts or an empty block of thoughts (just newlines or whitespace) we show an empty thought block

Are there known steps to reproduce?

Using qwen3, ask a single query and see that the reasoning is rendered correctly.
Now, update the workspace system prompt to start with /no_think and the rest of the prompt.
Sending a prompt will show an empty thought block.