A compact, document-focused specialist that parses and extracts structured content from PDFs, tables, and images. At 1.2B parameters it stays lean, trading broad general knowledge for precision on layout understanding and document parsing tasks. Expect reliable extraction work rather than open-ended reasoning.