I have fully configured all AI settings in the backend, including the chat section and spam detection (with model support for image input).
However, users are encountering a 500 error when using the title generation tool.
The proofreading tool for the same article works without issues. When I checked, it seemed there were no detailed logs? (Perhaps I missed them.) Screenshots are attached below.
You can query the table ai_api_audit_logs for detailed logs.
The 500 error is coming from Discourse, not from the Qwen API. So you need to remove the where clause on your SQL.
One easy way is triggering a title suggestion, getting the error, and then inspecting the newest line on that table.
Okay, thanks. Does the current AI support image generation, or is it limited to text?
For image generation see Improved Image Generation Support in Discourse AI
