OCR vs VLM

OCR extracts text from images. VLMs go further. What can VLMs do that OCR cannot? Flag format: CONGRESS{understands_[what]} Example: CONGRESS{understands_font_styles}