Compress PDF
to 50 KB

or drop one here
Compression level
Target size
Choose a PDF file first.

50 KB is 51,200 bytes, room for one lean page: a text export, a small greyscale scan, a signature on white. This page attempts the squeeze in your browser, hands back the smallest file it could prove, and tells you the number it actually reached.

The reference letter and the legacy box

Stefan is registering with a professional body whose portal has not changed in a decade: one PDF per proof, 50 KB each. His proof is a reference letter, reference-stefan.pdf, a single text page exported from a word processor at 84 KB, heavy with two full embedded fonts.

One run here drops it to 38 KB. Font subsetting alone does the job, cutting both fonts to the hundred odd glyphs the letter draws, and no pixel changes because no image was ever there. The upload passes, and the portal never knows how close it came to winning.

That is the file this number wants: short, textual, font heavy. Scans and photos are a different conversation, covered honestly below.

Attempting the 50 KB squeeze

  1. Open the page

    Press Choose a PDF, or drag the file into the tab. One page files open instantly and run in seconds.

  2. Confirm 50 KB

    This page presets 50 KB. If your box is roomier, type the real number instead of fighting this one for sport.

  3. Press Compress

    All four passes race locally. On text files the font pass usually wins outright; on scans the image pass fights for every byte.

  4. Read the honest number

    The result shows what the file reached, beside a preview. Press Download PDF only when the text reads well; a miss here is information, not failure.

How it works shows what each pass costs. When the file needs air, 100 KB doubles the budget in one click.

What 50 KB honestly holds

One text page, comfortably, when fonts do the shrinking. One small greyscale scan, where print reads but never crisply. One signature image on white. That is close to the full menu: 51,200 bytes split over two pages leaves 25 KB of image each, below the point where body text survives.

Photographs do not belong here, and neither do colour scans past a page. Colour triples the starting weight of a text page before compression begins, so a greyscale rescan at 200 dpi is routinely worth more than any setting. And every PDF carries a fixed structural cost of a few kilobytes that no pass can remove, which is why halving the cap collapses the odds instead of merely denting them.

Check what the box really means, too: some 50 KB caps are per page rather than per file, which changes everything. Size breakdown shows where your bytes sit before you commit.

Misses are the point of this page

Most files miss 50 KB, and the miss is useful: stopping at 62 KB means one small change away, while stopping at 400 KB means the document must change shape. Either way you receive the smallest file the engine could prove, with failed candidates discarded along the way and sub 98% text results refused.

Do not convert the pages to bare images to force the number; that destroys selectable text the checks here refuse to lose. Better levers: rescan in greyscale, drop to one page with Split PDF, or accept 100 KB and ask the recipient whether the box bends.

Better than worshipping the number

A box this tight is often negotiable in practice: a helpdesk ticket naming the cap and the page count resolves more cases than a twentieth run. A recipient who only needs to read the page once is happy with a photo of it in chat. And a file that must stay beautiful belongs at a humane target, not starved here.

Frank 50 KB answers

Can any PDF reach 50 KB?
No. Short text files and small greyscale scans can; photos and multi page scans generally cannot. The run reports the floor it hit, which is the honest answer for your file.
Why did my two page scan fail?
Two pages split 51,200 bytes into halves too thin for print. Greyscale rescanning helps more than squeezing; one page per file helps most.
What is the smallest a PDF can be?
A few kilobytes of structure plus whatever the content needs. Lean text files bottom out low; scans bottom out at readability, not at zero.
Should I strip the text to save bytes?
Never for a document someone must read, search or quote. Selectable text is the point of the format; the checks here defend it.
Is my file uploaded for this?
No. Everything runs in the tab, from your disk to your downloads. No account, no watermark, no queue.

Tools for tiny boxes

Split proofs apart, inspect the weight, or export pictures instead. Tab local throughout, verified before download.