PDF compressor

Compression results depend almost entirely on what's inside the PDF:

• Image-heavy PDFs — scanned documents, marketing brochures, photo books, product catalogs: typically shrink by 50–80%.
• Mixed PDFs with embedded photos — reports with screenshots: usually 20–50% smaller.
• Pure text / vector PDFs — invoices, ebooks, slides exported from Office: under 10% savings. Already optimal.

If your file is mostly text and you're not seeing meaningful savings, that's expected behaviour and not a bug — there's simply nothing left to compress without rewriting fonts or vector data, which would risk visual changes.

The three presets map to JPEG quality levels for the images embedded in your PDF. Text glyphs, line art and vector graphics are never re-encoded — they pass through untouched.

If none of the presets fit, open the Advanced panel and use the custom JPEG quality slider for fine-grained control between 20% and 95%.

Click Advanced… next to the quality picker to reveal the optional tweaks:

• Custom JPEG quality slider — set any value between 20% and 95% when the three presets don't quite hit the spot. Lower values shrink more but introduce visible compression artefacts.
• Strip metadata (on by default) — removes the embedded author, title, producer and creator properties from the output PDF. This is on by default for privacy: PDFs created by Word, Adobe Acrobat, macOS Preview and many web exporters embed the original author's name and the software used. Uncheck the box if you specifically need to keep the document properties (for example, when re-publishing an authored document and you want the byline preserved).

Your selections are remembered for the rest of the session, but never sync to a server and never persist after the browser tab closes.

Drop multiple PDFs at once — the tool processes them in parallel and shows a per-file progress bar with the before/after size. When everything finishes, a Download all (.zip) button appears that bundles every compressed file into a single archive, zipped locally with no external service.

There's no hard cap on how many files you can drop at once, but very large batches will use proportionally more RAM — we recommend staying under 20 files at a time on a typical laptop and under 10 on a phone.

The file's bytes are read with the standard File API, passed to the worker via a transferable ArrayBuffer, and the resulting compressed bytes come back the same way. At no point does any byte of your PDF travel over the network.

There is no telemetry on file names, file sizes or compression ratios. The page itself uses standard analytics for traffic counting, but no per-file event is ever emitted.

A few PDF types deliberately fall outside this tool's scope:

• Password-protected PDFs — decrypt the file first (open in your PDF reader, re-save without a password), then compress.
• Digitally signed PDFs — re-encoding any byte invalidates the signature. Don't compress signed contracts unless you plan to re-sign.
• PDF/A archival files — the conformance level (PDF/A-1, /A-2, /A-3) is not preserved by this tool. Re-validate after compression if you need PDF/A conformance.
• Scanned PDFs that need OCR — this tool only shrinks; it doesn't extract or recognise text. Run OCR first if you need a searchable output.