Having been using Drive’s OCR capabilities for years for an in-house Gmail add-on, it’s become less and less usable in the recent years (especially since 2023).
The add-on analyizes from 1 to 3 PDF files, with 2 calls per file to the Drive service (v2), over a 10-30s span. One for insert, one for trash.
Nowadays, even when limiting myself to only 1 file, I nearly always encounter a GoogleJsonResponseException, saying API call to drive.files.insert failed with the following error “User rate limit exceeded for OCR”.
I’ve created a dedicated project for that app/add-on, to ensure quota visibility, but the only detail I get is my 0.05% quota usage. I’ve tried adding a 120ms (then 230ms) pause prior to any API call, as some research suggested this might help (it doesn’t).
I’ve asked the Workspace Support AI, who also said to throttle my request (human suport said “we can’t help”, as usual).
The error says it’s a “user rate limit”, but I’ve found no information/details about such a limit. Also, the errror occurence rate seems to depend both on the day and time. Based on the general execution time, I get the impression this may depend on Google Cloud’s general load, and that I’m hitting some sort of background limit.
I’ve read about people encountering this over the years, but there was never a clear solution.
Shall I just assume Google wants people to stop using their OCR, and have just slowly made it more and more unusable with increasingly restrictive quotas ?