Compress PDF
Shrink a PDF by recompressing its images while keeping text selectable, free and entirely in your browser
Drag a PDF file here or click to choose
🔒 All processing happens in your browser — your files are never uploaded to any server
How to use it
- Upload a PDFDrag the file onto the drop zone or click "Choose file". It is read in your browser and never uploaded.
- Read the estimateWe inspect the file first and tell you how much of it is actually compressible, so you know whether it is worth doing before you start.
- Pick a quality levelBalanced suits most documents. Smallest file trades visible image quality for size; Best quality keeps images crisp and saves less.
- Click "Compress PDF"You get the before and after sizes, and the compressed file downloads. Your original is left untouched.
Enjoying the tool? Help us keep it free forever — Support OMXHub ❤️
Frequently Asked Questions
Will my text stay selectable and searchable?
Yes. We only re-encode the images inside your PDF — text, links and vector graphics are never touched. Many browser-based compressors turn every page into a picture, which does shrink the file but leaves you unable to select, search or copy a single word. We decided that was too high a price.
Why did my file barely shrink?
Because it was already efficient. Almost all the weight in a large PDF is in its photographs and scans; a text-only document is already only a few kilobytes per page and there is simply nothing to squeeze. We check your file before you commit and tell you what to expect, rather than spinning for ten seconds and handing back the same thing.
Can this tool ever make my file bigger?
No. Every image is kept only if re-encoding it actually came out smaller, and if nothing could be improved we return your original file untouched. A compressor that hands back something larger has failed at the one thing it was asked to do.
Which images get compressed?
JPEG images, which is what scanners and phone cameras produce and where the weight almost always is. Images using other encodings — or carrying transparency — are skipped rather than guessed at, because converting those correctly means handling colour spaces and alpha, and getting it subtly wrong would corrupt the picture instead of shrinking it.