Why "Structured Output" is not available with image attachment?

API
1

I think the doc should be updated to mention that the “Structured Output” is not available in this case, attaching an image to the input message (not generating image). It throws this error:

Invalid response_format: \'response_format\' of type \'json_schema\' is not supported with image message content types.

Is there a workaround? should I call the model first without Structured Output to ask it to do the OCR, then attach the OCRed text to the second message with Structured Output?
Or should I send the image as base64 encoded (in that case, is it reliable)? → Should be outrageously expensive : )

2

i tested on the playground for chat completion. seem to be working.

3

Thank, so it does work for chat completion, but not for assistant/thread.

Here’s the playground screenshot for assistant/thread:

image
image2764×2318 408 KB