Document tools
Extract text from PDFs and images
Bring the words out of your documents. Reuse searchable PDF text or recognize English scans, then save a plain text File alongside your original.
PDF · JPG · PNG → TXT · Free single-source processing
English · 20 MB per source · 3 pages per PDF · 3 pages per batch. Each source creates a UTF-8 TXT file (up to 2 MiB). Existing PDF text is reused; other pages use local English OCR. Originals stay unchanged.
Before you extract
How is text extracted?
Searchable PDF pages reuse their existing text. Pages without meaningful text and JPG/PNG scans use local English Tesseract OCR. Page order is preserved. A searchable heading can cause a scanned page to be treated as already searchable.
What does the download contain?
One UTF-8 .txt File per source, with LF line endings, no BOM, and a form-feed between PDF pages. Blank pages remain valid. TXT preserves words and page order; tables, layout, handwriting and non-English recognition are outside V1. Review OCR results for accuracy.
Where do my files go?
Your original is preserved. Text is stored in the generated private File under your existing storage and retention settings. Document contents are not sent to an external OCR or AI provider, or included in progress, logs, analytics or webhooks. You choose when to share or delete the File.
What are the limits?
Sources must be static, unlocked PDFs or still JPG/PNG images up to 20 MiB. TXT output is limited to 2 MiB. Free supports one source and 3 PDF pages; Pro supports 10 sources, 10 pages per PDF and 30 total batch pages; Unlimited supports 20 sources, 20 pages per PDF and 60 batch pages. Images are limited to 16 million pixels and 6,000 pixels per side. Processing time and storage limits apply.
Can I extract several files?
Paid plans support independent batches. Each source has its own Job and TXT File. Keep successful results, retry failed items only, and download successful TXT Files individually or in a ZIP up to 50 MiB.
Can I use workflows or the API?
Use extract_text through the shared processing API and SDK, or as the final step in a saved workflow. HEIC can first convert to JPG; scanned PDFs can first become searchable. Current tools do not accept TXT as their next input.