PDF-native extraction vs. vision models for document processingpymupdf.io6 points·by akaybk·8 miesięcy temu·1 comments