Is compressing a PDF on PDFWrangle really private?

Yes. The file is processed in your browser tab and never sent to pdfwrangle.com or anyone else. You can confirm it with the offline test or the Network panel on this page.

Does PDF compression work offline?

Yes, in both modes. Once the page has fully loaded, you can disconnect and keep compressing, with or without Convert pages to images.

How is this different from online compressors that upload?

Most online compressors process your file on their servers. Smallpdf, for example, says it deletes files after one hour (Smallpdf Compress PDF, checked 2026-10-20). With PDFWrangle there's nothing to delete, because the file never reaches a server.

Is it safe for confidential documents?

The file stays on your device, which removes the server side of the risk. Your own device still matters: browser extensions with permission to read web pages can see what a page shows, so for very sensitive files, use a browser profile or private window without extensions.

Compress PDF Without Uploading – Check It Yourself

Compress in three steps

Add your PDF.
Pick a level, or type a target size.
Compress and download the smaller copy.

For all the compression options, see Compress PDF.

How compression runs in your browser

Reading the file. When you add a PDF, the browser reads it from your disk into the page's memory. Nothing is sent anywhere.
Finding the images. pdf-lib, an open-source JavaScript PDF library, opens the file and lists the images inside it. Large 8-bit color or grayscale images stored as JPEG or Flate data are candidates; CMYK and other unusual images are left as they are.
Recompressing them. Each image is decoded by the browser, drawn smaller onto an off-screen canvas, and re-encoded as JPEG by the browser's own image encoder. If the new version isn't smaller, the original image is kept. With a target size, PDFWrangle steps down through quality and scale settings until the file fits or can't get smaller.
Saving. pdf-lib writes the new PDF in memory, and the download comes from a blob: link, which points at your browser's memory rather than at a server.
Convert pages to images. This option uses pdf.js (pdfjs-dist), Mozilla's open-source PDF renderer, to draw each page onto a canvas. pdf-lib then builds a new PDF from those page images. pdf.js runs part of its work in a helper file, the pdf.js worker, which your browser downloads from pdfwrangle.com once, in the background, shortly after the page loads. It's kept in memory and reused for every file.

What your browser downloads

Everything you see in the Network panel is your browser downloading something from pdfwrangle.com, not sending your file to it. When a page loads, that includes the page itself, its JavaScript files under /_next/static/, fonts, the logo, and small page-data files for the links in the menu (their names end in .txt?_rsc=…), which the site prefetches so other pages open faster. All of these are GET requests, and none carries your file.

The compression code, including pdf-lib, arrives with the page. A moment after the page appears, it also downloads pdf.js and its worker (about 1.4 MB in total) in the background, even if you never tick Convert pages to images, so the option is ready when you need it. The worker is a file under /v/ whose name starts with pdf.worker. It's downloaded once per visit, and nothing is fetched again when you compress.

Verify it yourself

Test 1: compress with the internet switched off.

Open this page and wait until it has fully loaded, then give it a few more seconds so the background download of the PDF renderer can finish.
Disconnect: turn off Wi-Fi or unplug the cable, or choose Offline in the throttling menu of your browser's DevTools Network tab.
Add a PDF, compress it and download the result. Then tick Convert pages to images and compress another one. Both work, because no server is needed.

If you disconnect too early, before the page has finished loading, the tool may not appear or may report an error. That's a download that didn't reach you, not a blocked upload. Reconnect, reload the page and start again.

Test 2: watch the Network panel.

Open the tool page and wait until it has fully loaded.
Open your browser's developer tools: press F12 (Windows, Linux) or Cmd+Option+I (Mac) in Chrome, Edge or Firefox, and select the Network tab.
Clear the list (the clear button in the Network toolbar), so only new requests appear from now on.
Add a PDF and click Compress. Try it with Convert pages to images ticked, too.
Look through the new requests. To show the Method column in Chrome or Edge, right-click a column header and tick Method. An upload would show up as a POST or PUT request with a body about as large as your file. You should find none. The page already downloaded the PDF renderer in the background while it loaded, so compressing usually adds no new requests to pdfwrangle.com at all. With Convert pages to images you may see a blob: entry: that's the pdf.js worker starting from your browser's own memory, not a transfer.

We ran both tests on the live site in Chrome on 20 October 2026, with our own test files. While the tools worked, the browser sent no POST or PUT requests and no request with a body. The only traffic was the GET downloads described on this page.

Compress PDF without uploading

Compress in three steps

How compression runs in your browser

What your browser downloads

Verify it yourself

Related tools

Frequently asked questions