What this tool does
Compressing a PDF does two things: rebuilding its internal object streams, and — when needed — re-encoding embedded JPEG images.
Everything runs in your browser. Files are never uploaded, and page count, page size and the selectable text layer are preserved.
Text-only PDFs shrink only a little. Scanned documents or photo-heavy PDFs often lose more than half their size.
How to use
- Drop your PDF into the area above One PDF at a time, up to 100MB. The file is opened locally and never leaves your device.
- Pick a compression level Lossless rebuilds structure only. Balanced resamples embedded JPEGs to 1600px longest edge at quality 80; aggressive uses 1000px at quality 60.
- Download and check the result You will see a before/after size comparison. Open the file to confirm text is still selectable and images are acceptable.
FAQ
Is my file uploaded anywhere?
No. All processing happens inside your browser; the server only serves the page. You can verify this in your devtools network panel.
Can I still copy and search the text?
Yes. Pages are never rasterized and fonts are not subsetted, so the text layer and page structure are preserved.
Why did my text-only PDF barely shrink?
In text PDFs the size lives in fonts and structure, and lossless compression does not touch fonts. Such files usually only shrink by removing pages or embedded fonts, which this tool does not do.
Can you compress a password-protected PDF?
No. Password-protected files are rejected outright; this tool never attempts to crack passwords. Remove the protection first.
What is the size limit, and what about larger files?
The limit is 100MB per file. For larger files, split them into parts first — browser tabs have limited memory and very large files tend to fail.
How different are balanced and aggressive?
Balanced is 1600px at quality 80 and suits most cases. Aggressive is 1000px at quality 60: much smaller, but photo detail visibly degrades — check it before printing.
Limits
Up to 100MB per file, one file at a time.
No decryption, no OCR, no font subsetting, and no downsampling of FlateDecode (lossless bitmap) images.
Compression is one-way: keep your original and re-export from it when you need maximum quality.