Split PDF
Share tool
Only this public address is included. Your files, inputs and results stay out of the link.
Extract the pages you need or divide a PDF into separate documents. Set page groups manually, by page count or with reviewed text-based suggestions.
Preparing your tool…
A visual example
One source, three complete groups
Example: split five pages after pages 2 and 4.
- 12345
Original PDF
The source has five physical pages.
- 1–23–45
Reviewed groups
Each page belongs to exactly one output PDF, in its original order.
The group export includes a manifest that maps source pages to their output files.
Illustrative example. This shows group mode. The separate extraction mode creates one PDF from selected pages. Text-based grouping uses existing PDF text, not OCR or an understanding of invoices.
How to use it
- Choose Review groups and export ZIP for a complete PDF batch, or select the existing Extract chosen pages into one PDF mode.
- For groups, choose a fixed page count, explicit start pages or a literal text anchor within a confirmed page region. Review missing/ambiguous values and repeated headings.
- Check every group and safe unique filename; use {index}, {start}, {end} or {text}, or edit names individually. Confirm the complete page coverage.
- Export the separate PDFs in a ZIP with manifest.json, which records every original page and its output file.
The five-page invoice example repeats INV-001 on pages 1–2 and INV-002 on 3–4. The text-region workflow suggests groups 1–2, 3–4 and 5 without splitting each repeated heading.
Details & limits
Groups: one PDF up to 25 MiB / 100 pages, with complete verified page coverage. Legacy page extraction: 20 MiB / 200 pages. Neither is a PDF security-cleaning service.
Can the grouping workflow drop pages or automatically understand scanned invoices?
No. Group export requires every source page exactly once in its original order, and reopens each PDF to check page counts, dimensions and rotation. Text matching is a bounded literal anchor in existing PDF text, not OCR or semantic invoice recognition; confirm missing/ambiguous matches and names. Up to 25 MiB/100 pages, 64 MiB output. Forms, signature fields and encryption are unsupported. The separate page-extraction mode still accepts ordered ranges such as 3, 1-2 and creates one PDF, rejecting duplicate, reversed or out-of-range pages. Neither mode is security redaction.