Hacker News new | ask | show | jobs
by incomingpain 303 days ago
Obviously all the big named online AI will do this trivially. Upload file, ask for everything you want in whatever format you want. If I were doing it: https://mistral.ai/news/mistral-ocr

To do it offline due to privacy, vision enabled LLM. Biggest Gemma you can handle, qwen2.5 vl, or Mistral small. I'd probably choose mistral.

Openwebui does pdfs built in. https://docs.openwebui.com/features/document-extraction/

TBH havent tried it myself but I bet it works.