Compress PDF

Shrink a PDF two ways — lossless re-save, or smallest size as page images.

Runs in your browser. Nothing is uploaded — your input is processed on this device and discarded when you close the tab.

Your file

Drop a PDF here or click to browse. One file at a time, up to 50 MB. Everything is processed on this device only — no file is uploaded.

Compression mode Keep text — lossless re-save; text stays selectable, gains are modest Smallest size — pages become JPEG pictures; text is no longer selectable or searchable

Read the mode descriptions on this page before choosing “Smallest size” — that conversion cannot be undone from the output file.

Raster resolution for “Smallest size” (DPI) 100 DPI — smallest, softest 150 DPI — balanced 200 DPI — sharper, larger

JPEG quality for “Smallest size” (0.5–0.9)

Compress PDF

Every PDF eventually meets a size limit: the email attachment cap, the upload form that rejects anything over 5 MB, the portal that wants one file and wants it small. This tool shrinks PDFs in your browser, and it is unusual in one respect: it offers two modes and tells you plainly what each one costs. "Keep text" re-saves the document losslessly — the words stay words — and accepts whatever modest saving the file's structure allows. "Smallest size" converts every page into a JPEG picture and rebuilds the document from those pictures, which can shrink a scan-heavy file dramatically but destroys the text layer in the process. Nothing is uploaded in either mode; the work happens in your tab, which is why the file cap is 50 MB.

The choice between the modes is a real one, and the rest of this page exists to help you make it with your eyes open. A compressed file that nobody can search, copy from or read with a screen reader is not always a bargain, and a lossless re-save that saves almost nothing is not a failure — it is information about how tidy the original already was. The tool reports the true before-and-after byte counts either way, so you can judge the result rather than a promise.

How to use Compress PDF

1. Add your PDF. Drop it onto the workspace or click to browse. One file at a time, up to 50 MB. The tool reads the page count immediately, and a damaged or password-protected file is reported before anything runs.
2. Read the mode descriptions, then choose. "Keep text" if the document must stay searchable and selectable — contracts, reports, anything you copy figures out of. "Smallest size" if it only needs to be looked at or printed and the file size is the problem — typically scans and photo-heavy documents.
3. For "Smallest size", set the resolution. 150 DPI is the balanced default: sharp enough to read comfortably on screen. Drop to 100 DPI for the smallest file when the pages will only be viewed small; rise to 200 DPI when fine print must survive.
4. Set the JPEG quality. The slider runs from 0.5 to 0.9, defaulting to 0.75. Lower means smaller and softer; higher means sharper and larger. Text pages tolerate the low end surprisingly well; photographs with gradients show artefacts first.
5. Compress and read the result line. The status line gives the exact sizes — for example how many bytes went in and came out, and the percentage saved. In "Smallest size" mode it also repeats the warning that the pages are now pictures.
6. Verify before you rely on it. In "Keep text" mode, open the result and copy a line of text out of it. In "Smallest size" mode, zoom into the densest page and confirm the fine print is still legible at the DPI you chose. Keep the original until you have done this.

Features

• Two clearly separated modes: lossless re-save ("Keep text") and rasterise-and-rebuild ("Smallest size").
• "Keep text" re-saves with object streams enabled, packing the file structure tightly; text, links, form fields and vector content are preserved exactly as they were.
• "Smallest size" rasterises at 100, 150 or 200 DPI with JPEG quality from 0.5 to 0.9, and rebuilds each page at its original point size, so the document prints at the same physical dimensions.
• Page count and page sizes are preserved in both modes.
• Honest reporting: the exact input and output byte counts and the percentage change are always shown, including when the answer is "no smaller".
• Processing runs entirely in the browser using the pdf-lib and pdf.js libraries; the original file on your device is never changed.

Worked example

The figures here are from this tool's own automated test runs, not estimates. The test fixture is a four-page document: three text pages plus one page carrying a 600 × 600-pixel noisy image at full page size — the kind of page that makes scans heavy. It measures 1,082,307 bytes. In "Smallest size" mode at 100 DPI and quality 0.5, the output measures 256,063 bytes — 76.3% smaller, still with four pages at their original sizes, and text extraction from the result returns nothing at all, confirming the pages are pictures. The same class of file through "Keep text" came back at 1,082,307 bytes from 1,082,305 — two bytes larger, a true zero — because pdf-lib had written the fixture tidily in the first place and there was no structural slack to recover. And a warning in the other direction: a small 1,617-byte text-only file rasterised at 150 DPI came out at 42,485 bytes, over twenty-five times larger. Rasterising a document that is already mostly text is a way to make it bigger, not smaller.

Common errors and limitations

• "Smallest size" destroys the text layer. Pages become JPEG images: text cannot be selected, searched, copied or read aloud by a screen reader; links and bookmarks stop working; form fields become pictures of form fields. This is stated in the mode picker, here, and again in the result line, because it is the single most important fact about the mode.
• "Keep text" cannot shrink image data. A PDF whose weight is photographs or scans will barely move in this mode, because lossless re-saving only repacks structure — it does not re-encode images.
• Neither mode performs OCR: a scanned page stays a picture in both modes ("Keep text" preserves it exactly; "Smallest size" re-photographs it).
• Rasterising long documents takes time and memory in the tab — every page is rendered and encoded in sequence. Very long or very large documents may be slow, and the 50 MB cap still applies.
• Password-protected PDFs cannot be opened in either mode; remove the password in the app that created the file first.
• One document at a time. There is no batch queue.

• Using "Smallest size" on a document people need to search. The output looks identical and weighs less — and every word in it is now paint. Fix: use "Keep text" for anything searchable, and reserve "Smallest size" for look-only documents such as scans of signed pages.
• Expecting "Keep text" to shrink a photo album. Its gains come from file structure, not image re-encoding; a tidy, image-heavy file can come back the same size or a few bytes larger. Fix: that result is the honest answer for a lossless pass — if the size must fall, the image mode is the tool, with its costs understood.
• Choosing 200 DPI and quality 0.9, then wondering why the file grew. High resolution and high quality can exceed the original's embedded images, especially for born-digital text files. Fix: start at 150 DPI and 0.75, and only raise settings for pages whose fine print genuinely needs them.
• Deleting the original after compressing. Especially in "Smallest size" mode, the conversion cannot be undone from the output. Fix: keep the original until the compressed copy has been opened, read and searched (or consciously accepted as unsearchable).

Frequently asked questions

Related PDF tools

Prefer everything on one page? The original PDF Toolkit merges, splits and rearranges with one shared file list. The full suite is indexed on the PDF Tools hub.

How it works

Compress PDF shrinks a PDF in your browser using one of two modes, and shows real before-and-after byte counts so you can judge the result yourself. "Keep text" performs a lossless re-save: it repacks the file's structure while keeping the text selectable and searchable, along with links and bookmarks where they exist. "Smallest size" rebuilds each page as an image, which usually produces a much smaller file but turns every page into a picture. The tool never uploads your document, and it reports the actual size change rather than a rounded claim. Choosing between the modes is the main decision: keep the text layer, or accept pictures in exchange for the smallest file. Always check the downloaded file before sending it on, opening a middle page as well as the first, so a mode that looked right in the summary does not surprise you in the document itself.

Another worked example

A fictional office worker, Sam Taylor, has a 4.2 MB meeting report that is too large for an email attachment limit. Sam first tries "Keep text". Input: a 4.2 MB report with selectable text. Result: a 3.9 MB file with text intact — barely smaller, because the bulk is photographs. Sam then tries "Smallest size" on a copy and gets a 1.1 MB file, but the text can no longer be selected or searched. Sam sends the "Keep text" version by a file-sharing link instead, and keeps the smallest version only as a last resort for a plain reading copy.

More limitations to know

"Keep text" cannot shrink image data: a PDF whose weight is photographs or scans will barely move in that mode, because lossless re-saving only repacks structure and does not re-encode images. "Smallest size" destroys the text layer: text cannot be selected, searched, copied, or read aloud by a screen reader, and links, bookmarks, and form fields stop working. Neither mode performs OCR, so a scanned page stays a picture in both modes. If a file is already tightly packed, either mode may save little, and the honest byte counts will show that.

More frequently asked questions

Which mode should I choose?

Choose "Keep text" when the document must stay searchable and its links and bookmarks matter. Choose "Smallest size" only when file size matters more than selectable text.

Why did "Keep text" barely shrink my file?

Its bulk is probably photographs or scans. Lossless re-saving repacks structure; it does not re-encode image data.

Will the smallest version still be searchable?

No. Its pages are images, so text cannot be selected, searched, or copied.

Does compression upload my file?

No. Both modes run in your browser.

Related articles