Sometimes you intentionally want to verbatim keep "mistakes", sometimes you don't and want them to be "fixed". OCR-only models tend to only do one of those two,

Sometimes you intentionally want to verbatim keep "mistakes", sometimes you don't and want them to be "fixed". OCR-only models tend to only do one of those two, in VLM cases often the latter. With multi-modal LLMs you can just tell them (adherence of course needing evals/differs per model).