Shrink a PDF. The lossless pass rewrites the file structure and keeps text selectable; the image pass rebuilds every page as a JPEG for scans. Nothing is uploaded โ the file stays in the browser.
Drop a PDF here, or click to choose
One file, up to 50 MB. It is processed in your browser.
When a PDF will not shrink, the file usually has nothing left to squeeze: fonts are already subset, images are already JPEG, objects are already packed. So this page offers two passes โ decide what you need to keep. Keep the text, take lossless. Just need it small, take the image rebuild.
Both run in the browser: lossless rewrites the file with pdf-lib, the image pass renders each page with pdf.js and embeds it into a fresh PDF. Nothing is uploaded, and closing the tab leaves nothing behind.
The document is read and written out again: objects orphaned by earlier incremental saves are dropped, everything is packed into object streams, and the document information dictionary (title, author, subject, keywords, producing app) is cleared. Text stays text, links still click, bookmarks survive โ the recipient can copy and search as before.
Every page is rendered at the resolution you pick, encoded as JPEG and stitched into a new PDF. The physical page size is preserved (A4 stays A4), but each page now contains a single image: no selection, no search, visible blocks past 200% zoom.
Three usual reasons: the images are already compressed JPEGs, the fonts are already subset, or the file is mostly text with few images. None of those have slack โ a text-only PDF is often a few hundred KB and does not need compressing. The files that shrink dramatically are almost always scans wrapped in a PDF shell.
One file at a time, up to 50 MB. A password-protected PDF is rejected at read time, because pdf-lib implements no /Encrypt handling. The whole job runs in browser memory โ there is no endpoint that accepts a PDF, and closing the tab discards everything.