Compress PDF

The file is rewritten with object streams — many small PDF objects packed into a few compressed streams — and objects nothing references any more are left out. Optionally the document metadata (title, author, dates, XMP packet), form fields (flattened into page content) and embedded file attachments are removed. The before and after sizes are the real byte counts of your file and the rewritten one; when the result is not smaller, no download is offered.

Images are not re-encoded or downsampled and fonts are not subset. Those are where most of a scanned or photo-heavy PDF's size lives, so such files shrink little here — a real image-recompressing tool (or re-exporting from the source application at lower image quality) is the right route for them. The page says this plainly rather than promising a percentage.

Flattened forms can no longer be filled; removed metadata is gone from the copy (your original is untouched). Encrypted PDFs are refused. The reading and writing is done by pdf-lib, an open-source (MIT) PDF library that is downloaded to your browser only when you choose a file — the page itself stays small. Nothing leaves the browser: the file is read into memory on your device, rewritten there and offered back as a download; the same four anonymous usage counts as the rest of the site apply.