How to extract text from PDF with Mistral AI?

,

Hello,

Does anybody know how to use OCR Mistral AI endpoint ?
Using their documentation, the response is still the same : 422 empty error message

Welcome to the Make community!

The document format or URL might be unreadable by Mistral. Try another method like file upload, or another file hosting application.

Hope this helps! If you are still having trouble, please provide more details.

@samliew
P.S.: investing some effort into the tutorials in the Make Academy will save you lots of time and frustration using Make!

Hello samliew,

Here is the link : https://drive.google.com/uc?export=download&id=19HCyGZ5zkcEPFfFgb9lgwlhW5THo2qav

There is no restriction on the download. I even tried on n8n on the mistral completion module with the same model and it worked. I don’t figure out why I meet this error on make.com since it also already works like this for other members of the community here.

I tried to store the file on mistral ai thanks to their endpoint, but I meet the same empty error again.

For the OCR issue, I found out.
Simple mistake : documentUrl (:cross_mark:) => document_url (:white_check_mark:).

For the second case : impossible to use application/json with base64 content of the file. It seems that only multipart/form-data would work, thus it requires to convert the pdf file into jsonl.

1 Like