Compress PDF on Mac

The built-in Quartz filter applies one aggressive, non-configurable downsampling profile designed years ago for email attachments. It cannot target a size, and on documents that are already optimized it can actually increase the file size because it re-encodes content that did not need touching.

No. Everything runs inside your browser tab. There is no .dmg, no Homebrew formula, no command-line Ghostscript setup and no admin password prompt — which also makes this usable on managed corporate Macs.

Yes, and it is noticeably fast there. WebAssembly compiles to native ARM64 instructions on M-series chips, so compression on an M1 or later typically runs several times quicker than on comparable Intel hardware.

Compression is one-way, but your original file is never modified — it stays exactly where it was on your device. The compressed version is written as a separate download, so you always retain the source and can simply try again with a gentler preset.

No. Compression runs entirely inside your browser using WebAssembly. Your PDF is read into local memory, processed on your own CPU and written straight back to your downloads folder. No network request ever carries the file.

Yes. Bookmarks, the table of contents, internal cross-references, hyperlink annotations and interactive form fields all survive compression. Only the encoding of images and content streams is altered — the document's structural and navigational layers are preserved.

Yes, with no file limits, no watermarks and no account. Because processing happens on your device rather than our servers, there is no per-file cost to recover and therefore no reason to meter your usage.

Progress tracks pages completed, not time elapsed, and pages differ enormously in cost. A page holding one large photograph takes far longer than ten pages of plain text, so the bar appears to stall. It is still working — the Web Worker keeps the interface responsive throughout.

It depends on the content. Scanned or image-heavy documents commonly shrink by 70–95%. Text-only PDFs that were already optimized may only drop 5–20%, since there is little redundant data left to remove.

There is no page-count limit. The practical constraint is memory rather than pages: a 500-page text document compresses easily, while a 500-page high-resolution scan may exhaust memory on a lower-spec machine. Closing other tabs frees up headroom for very large jobs.