The extension works well and the integration looks good, except that all of the responses I get from the vision model go straight to the console while the UI reports:
Vision Model Responses:
None
The responses that go to the console are good, and occasionally I'll get two responses, a quickly returned short one in the UI which seems suspicious followed by a longer wait and then the "None" response with the actual response showing up in the console. Any ideas?