Convert PDF, Word, PowerPoint, Excel, CSV and web pages to Markdown with headings and tables kept, or heading-aware chunks for a vector database. Each row has title, markdown, pageCount, wordCount, tokensApprox and needsOcr. Mixed formats in one run; failed files free. $1 per 1,000 pages.
All dates are UTC. Build numbers are the Apify build that shipped the change.
2026-11-14 (0.1.21-0.1.22)
Page range: convert, and pay for, only the PDF pages or slides you list (1-5, 9, 12-). Page markers keep the real page numbers.
PDF password: password-protected PDFs open when you give the password (stored encrypted). Without it, or with a wrong one, the file fails free with a message saying which.
Documents at a time (maxConcurrency, 1 to 10, default 3) in the form.
New author and createdAt fields on document rows, from the file's own properties. Placeholder names and email addresses are left out.
Files can be sent inside the input as base64Files (data URI, plain base64 or {fileName, data}), for API and agent callers.
Hidden field names for API callers: urls, url, pdfUrls, maxPages, maxItems, pdfPassword, base64Pdfs. Existing inputs work as before.
Dropbox /s/ and /scl/fi/ share links without dl= now download the file.
README: question headings, a section on pages, passwords and direct files, and measured runs.
2026-11-05
Store copy rewritten
2026-10-16 (0.1.15-0.1.17)
README updates only (rating request, links tidied). Output unchanged.
2026-10-15 (0.1.12-0.1.14)
Memory guard: when a very large file is about to exhaust the run's memory, that document is stopped (not charged) and the run carries on with the rest. Minimum run memory is 512 MB.
A key-value store that doesn't exist or can't be read now ends the run with a clear message and no charge, and a mistyped store name no longer creates an empty store in your account.
Size limit for a document row is measured in bytes, so long documents in Chinese, Arabic or other multi-byte scripts are cut cleanly instead of failing the push.
2026-10-14
Store description now states the price per 1,000 pages.
2026-10-13 (0.1.10-0.1.11)
An arXiv abs link and its pdf link to the same paper are converted and charged once.
Each document's timeout is cut to the run's own time limit, so a run that runs out of time finishes cleanly and lists the rest as skipped (not charged).
Store title, description and README opening rewritten.
2026-10-11 to 2026-10-12 (0.1.7-0.1.9)
Public release on the Apify Store on 2026-10-11, plus further builds (details not recorded).
2026-10-10 (0.1.1-0.1.6)
First release: PDF, DOCX, PPTX, XLSX, CSV, HTML, Markdown and text to Markdown, one row per document or heading-aware chunks with page numbers, scanned PDFs flagged with needsOcr, failed documents free.