Files stay on your device 4 presets 20 MB per file

PDF Compressor

Scanned documents, presentation exports and e-tickets are usually far bigger than they need to be. Pick a preset, watch the size drop, and download — the file is rebuilt by your browser, not by a server somewhere else.

PDF workspace Waiting for a PDF Local only

Drop a PDF here, or click to browse

Up to 20 MB. Everything is processed in this tab.

PDF
Renders each page at 150 DPI and re-encodes it as JPEG. Best balance for scanned documents and image-heavy PDFs.
Works offline once the page has loaded.

Contracts, invoices, medical reports

The PDFs people most want to shrink are usually the most sensitive ones. Here the renderer and the PDF writer both run inside your browser, so the document is never transmitted — which also means no upload wait, even on a slow connection.

How browser-based PDF compression works

A PDF is a container: pages, fonts, vector graphics and images packed together, often with redundant metadata and unused objects. Shrinking one is a trade-off between three things — file size, visual quality and whether the text stays machine-readable.

Lossless cleanup (no quality loss)

This rebuilds the document with a fresh writer, drops XMP metadata, empties author and title fields, recompresses the file structure using object streams and discards unused objects. Text remains selectable, links keep working, and nothing is re-rendered. Real-world savings are typically 2–15%, occasionally much more on files exported by older tools that left a lot of junk behind.

Raster compression (the big win)

The other presets render every page at a chosen DPI and re-encode it as JPEG. This is exactly how a photocopier-style compressor works, and it is what makes an 18 MB scan become 1.5 MB. The trade-off is real: the page becomes a picture, so text is no longer selectable, searchable or accessible to screen readers, and hyperlinks inside the document stop working.

Rule of thumb:

  • Scanned paperwork, photos, flyers → Recommended or Strong. You will rarely notice the difference on screen.
  • Text documents you still need to search → Lossless. Keep the text layer intact.
  • Email attachments with a hard size limit → Maximum compression, optionally grayscale.

What about those “compress to 90% smaller” sites?

Most of them do exactly what the Recommended preset does — rasterise the pages — while presenting it as magic. The honest version is on this page: you choose the DPI, and you can see the quality trade-off in the result.

Limits and edge cases

  • Password-protected PDFs cannot be opened. Remove the password first.
  • Digitally signed PDFs lose their signature validity when the file is rebuilt — keep the original for archival.
  • Very large documents (hundreds of pages at high DPI) use a lot of memory; close other tabs if the browser struggles.
  • Already-optimised PDFs may only shrink a little. If the result is bigger, the tool says so instead of pretending to win.

Need to build a PDF from images instead? Use the JPG/PNG/WEBP to PDF converter — its “Small” quality setting produces deliberately compact files from the start.

Frequently asked questions

Is my PDF uploaded?
No. The PDF is read with the File API and processed by JavaScript running in this tab. There is no upload request — you can confirm in your browser’s Network panel.
Which preset should I pick?
Start with Recommended (150 DPI). If it is still too big, move to Strong or Maximum. Use Lossless whenever the text must remain selectable or searchable.
Will my PDF still look right when printed?
Page dimensions in millimetres are preserved exactly, so printing uses the same paper size. At 150 DPI, detail is fine for normal reading; for fine print or technical drawings, keep lossless or use a higher DPI.
Why did my file only shrink by 3%?
The PDF was probably already optimised (many exporters compress on save). In that case only re-rendering — the Recommended preset or lower — will make a big difference.
Does it remove metadata and hidden info?
Yes. Author, title, subject and keyword fields are cleared in every preset, and rasterising removes embedded scripts, forms and annotations along with the original object tree.