Google vision API returns empty text for the same image document after the recent changes on December5 2023 https://cloud.google.com/vision/docs/release-notes.
Previously version of Vision API correctly returned the text from the document, however the new API returns blank OCR text.
If I use older model with builtin/legacy option, I still get the correct results. However this older model will be deprecated soon and I am trying to find alternate solution for this before that.
Pretty sure you are supposed to fill a report to GCP bug tracker here: https://issuetracker.google.com/issues/new?component=187174&template=0&pli=1 , attaching the document that their service stopped processing correctly. They are supposed to fix the issue -- that's what the API version migration period is for.