Process documents with hundreds of pages with no issues. In this example, I extracted crypto holdings from 200+ page SEC filings by first classifying pages using VLM support and then extracting relevant information only from those pages.