Compress PDF
Make a PDF smaller without handing it to anyone. The file is opened and rewritten inside this browser tab, so it never reaches a server — ours or anybody else’s. Part of the Switch Labs PDF toolkit.
Drop a PDF here
The file is opened in this browser tab and never uploaded. Open your network tab and watch — there is no request carrying your document.
What actually makes a PDF large
Almost every oversized PDF is oversized for one of three reasons, and knowing which one tells you whether compression will help at all.
Scanned pages (usually the culprit)
A scanner produces one large photograph per page, often at 300 or 600 DPI, and often stored with little compression. A twenty-page scan can easily be 40 MB. This is the case where re-rendering wins enormously, because the pages are already images and re-encoding them at a sane resolution loses very little that you can see on a screen.
Embedded photographs and screenshots
A slide deck exported to PDF carries every image at full resolution, whether or not it is displayed at that size. Same fix, same large payoff.
Embedded fonts and metadata
A few hundred kilobytes at most. The structural method strips metadata and repacks objects, which is worth doing and is not going to solve a 40 MB problem.
When compression will not help: a text-only PDF is already close to as small as it can be. If yours barely shrinks, that is the honest answer, not a failure of the tool. Look instead at whether you need every page — splitting it is often the real fix.
Why on-device matters more here than anywhere else
Think about what people actually compress. Signed contracts. Medical records. Tax returns. Court filings. Passport scans for a visa application. It is close to the highest-sensitivity file category a normal person handles, and the standard way to compress one is to upload it to a website you have never heard of, in exchange for a promise that it will be deleted in an hour.
This tool cannot make that promise because it never needs to. There is no upload step. pdf.js renders the pages, canvas re-encodes them, pdf-lib assembles the result, and all three run as JavaScript inside the page you are looking at. That claim is checkable in about ten seconds with your browser’s network tab open, which is the only kind of privacy claim worth making.
The practical cost is real and we would rather name it: browser memory is finite, so very large documents fail here and would succeed on a server. That is the tradeoff. For the files people are most nervous about, it is the right one.
Which method to use
| Your document | Use | Why |
|---|---|---|
| Scanned contract or receipt | Re-render, eBook preset | Already images; nothing to lose, a lot to gain |
| Slide deck exported to PDF | Re-render, eBook or Screen | Full-resolution images shown at slide size |
| Report you still need to search | Structural | Keeps text, links, and bookmarks intact |
| Form someone has to fill in | Structural | Re-rendering destroys form fields |
| Something going to a printer | Re-render, Print preset | Holds up on paper; still much smaller than a raw scan |
Frequently asked questions
Is my PDF uploaded to a server?
No. The file is read into this browser tab and processed there by pdf.js and pdf-lib, both running as JavaScript in the page. Nothing is transmitted, so there is no copy on a server, no retention window, and nothing for us to lose in a breach. You can verify it: open your browser's network tab and compress a file. No request carries the document.
How much smaller will my PDF get?
It depends entirely on what is inside it, and any tool quoting a single percentage is guessing. Scanned documents and image-heavy decks often fall by most of their size, because scans are photographs and photographs re-encode well. A text-only PDF may barely move, because there is nothing bulky in it. The tool reports the measured before and after byte counts rather than a projection.
Will my text still be selectable?
Only if you use the structural method. The re-render method converts each page into an image, which is what makes it so effective on scans, and images are not text. Copy, search, links, form fields, and accessibility tags do not survive it. The structural method leaves all of that intact but saves far less.
Does this run OCR or improve scan quality?
No on both counts. Compression only removes information; it never adds any. A blurry scan stays blurry, and text in an image stays an image. If you need scanned text to become real text, use an OCR tool first, then compress.
Can it open password-protected PDFs?
No. If a PDF is encrypted, remove the password in whatever reader you already use to open it, then compress the unlocked copy. The tool tells you plainly when a file is protected rather than failing with something cryptic.
Is there a file size limit?
About 100 MB, and 300 pages for the re-render method. Those are browser memory limits, not policy. Beyond them a tab starts thrashing, and failing with a clear message is better than hanging. For a very large document, split it first and compress the parts.
The rest of the PDF toolkit
- Split PDF — extract pages or divide a document
- Merge PDF — combine documents into one
- Image to PDF — JPG, PNG, WebP, or HEIC into a PDF
- PDF to Word — convert to an editable document
- Fill & sign — complete and sign before you compress
- How this compares to iLovePDF — including what we do not do
Related Tools
More free tools that might be useful
Split PDF
Extract pages or divide a PDF into separate files
Image to PDF
Turn JPG, PNG, WebP, or HEIC images into a single PDF
PDF Rotate, Reorder & Extract Pages
Rotate, reorder, extract, and delete PDF pages online
Combine Files
Merge multiple CSV or text files into one