My questions relate to the functionality mentioned here:
The public URL requirement is unusual. On the above documentation page, a password protected PDF is mentioned, which makes sense given it would be on a public server. I can’t find any mention on how to supply the PDF password to the chat completions API?
Also, I just wanted to verify that I would be able to use chat completion structured output and that the PDF evaluation would be multi-modal (text and images)?
So three questions:
-
How do PDF passwords work with the chat completions API?
-
Does the PDF evaluation by Sonar also analyze images?
-
I assume I will be able to use your structured output to get the results in JSON format?