Feed documents to your model as text, not pictures.
PDF, Word, PowerPoint, Excel, HTML, EPUB and more in — clean Markdown out, with headings, lists and tables the model understands natively. Cheaper tokens, better answers.
Free for simple conversions — no sign-up.
Convert
Drop a document here or click to choose — PDF, DOCX, PPTX, XLSX, HTML, EPUB… up to 10 MB
Engine
Scanned PDFs (OCR), complex tables, files up to 50 MB, URLs, history and an API key:
sign in. Nothing you convert here is stored.
Three engines, picked per document
markitdown Microsoft's converter for Office files, HTML, CSV/JSON/XML, EPUB and ZIP archives.
pymupdf Fast Markdown from text PDFs: headings by font size, ruled tables, reading order.
docling IBM's layout model with OCR (Portuguese + English) and table structure recognition, for scanned or table-heavy PDFs. Signed-in users.
What signing in adds
- Docling with OCR, files up to 50 MB and 600 pages, conversion by URL
- A queue for slow conversions and a history of your last documents (kept 30 days)
- Your own API keys —
curl -F file=@doc.pdf https://prensa.lucas.mat.br/api/v1/convert— and a remote MCP server for Claude Code
Anonymous conversions are limited per address and per day; documents are processed in memory and discarded. Built on MarkItDown, pymupdf4llm and Docling.