PDF

Merge, split, protect and convert — none of it uploaded.

All 15 PDF tools

Every one runs in your browser. Nothing is uploaded and nothing is metered.

What PDF tools are

Utilities that read and rewrite PDF documents — combining, splitting, reordering, encrypting, converting. Almost all of them work by uploading your file to a server; these do the same work in your browser, which removes the limits that exist to ration server time.

Fifteen PDF tools that run in your browser. Your document is never uploaded, so there is no queue, no two-files-a-day cap and no account.

Some conversions cannot honestly be done client-side. Those are listed below with the specific reason, rather than left for you to discover after choosing a file.

Why there is no upload

Every mainstream PDF service works the same way: you upload, a server processes, you download, and the file sits on someone else’s disk for an hour or a day. That architecture is what creates the limits — two files a day, twenty to a batch, a megabyte ceiling, an account before you can do it twice — because each operation costs the provider machine time.

These run on your processor. There is nothing to meter, so nothing is metered, and your contract or medical letter does not leave the device it is on. Verify it by opening the network tab, or by disconnecting entirely.

What makes these PDF tools different

  • No file count, because there is nothing to ration

    Each operation costs this site nothing, so there is no reason to cap it. The practical ceiling is your own memory, which for text documents runs into thousands of pages.

    iLovePDF allows 25 files per merge; Smallpdf gates batching behind a subscription; CloudConvert counts a merge against ten conversions a day.

  • Encrypted documents are decrypted, not worked around

    Enter the password once and the pages are decoded properly before anything is copied. Protection goes back on with AES-256, the cipher the current specification names, and RC4 is refused rather than offered as a compatibility option.

    The widely used open-source PDF library cannot decrypt at all: its flag for handling encrypted documents lets a broken file through rather than decoding it, and merging one produces blank pages with no error.

  • Compression that reports the real number

    What is permissively available is structural — repacking object tables, compressing streams, dropping unreferenced objects. On a document produced by current software that saves close to nothing, and on a very small one it can cost a few bytes. The figure is shown either way, including when it is negative.

    Genuine content-aware downsampling requires Ghostscript or MuPDF, both AGPL-3.0; shipping either to a browser would oblige this entire site to that licence.

  • The refusals are published, not discovered

    Six operations are named on the hub with the specific reason each one is absent, so you can go elsewhere before choosing a file rather than after uploading one.

    PDF to Word, Office to PDF, HTML to PDF, AI summarising, content-aware compression, and removing a password you do not have.

How to do it

The order that works, and why each step changes the answer.

  1. Work out which operation you actually need

    Reordering is Organise PDF, not a split. Hiding a margin note is not cropping. Naming the operation correctly is most of the work, and it is where the wrong-tool mistakes happen.

  2. Deal with encryption before anything else

    If the document asks for a password, unlock it first. Every later operation is simpler on a decrypted file, and a merger that papers over encryption produces blank pages rather than an error.

  3. Check the pages against the thumbnails

    Split states the pages it will produce before it runs, and Organise labels every thumbnail with its position in the output. Check those against the previews rather than trusting remembered page numbers.

  4. Leave the destructive step until last

    Deleting pages, stripping metadata and flattening are one-way. Run them at the end and on a copy, so an earlier mistake does not cost you the original.

  5. Confirm nothing left the device

    Open the network panel and watch the request count while you work, or disconnect entirely. A privacy claim you can check in ten seconds is worth more than one you have to take on trust.

Worked examples

Real inputs, real numbers. Pick one.

Cropping mistaken for redaction

You give it

A contract cropped to hide a note in the margin.

You get

The CropBox moves and the MediaBox does not, so the note is still in the file and any text extractor will return it. Page geometry is in PostScript points at 72 to the inch — an A4 page is 595 × 842 — so a 36pt trim hides half an inch of display area while deleting nothing.

Choosing a rendering resolution

You give it

One A4 page exported as an image at 150 dpi and again at 300 dpi.

You get

Pixel count scales with the square of the resolution, so 300 dpi produces a file roughly four times the size of 150 for the same page. 150 is the sensible default, 300 is for print or OCR, and 72 gives the page at its nominal size.

A compression run that saves almost nothing

You give it

A text-only PDF exported by current word-processing software.

You get

The structural work was already done by the producer, so the saving is close to zero, and on a very small document the object-stream overhead can exceed it. The result then shows a negative figure rather than hiding it, and the honest reading is to keep the original.

Which PDF tool do you need?

Start from the job, not the tool name.

Why the free tiers are capped

AspectdsyniLovePDF, Smallpdf and similar
Files per operationUnlimitedMerge 25, compress 2, split 1
Daily limitNoneCommonly 2 tasks a day without an account
File retentionNever receivedDeleted after 1–2 hours
Encrypted PDFsDecrypted properly with your passwordOften produces blank pages
CompressionLossless, or rasterise — the real figure reported either wayA percentage with no method stated
PDF to WordRefused, with the reasonOffered, quality varies

The figures behind that

  • AES-256

    is the cipher specified in ISO 32000-2; RC4, which some tools still default to, has been considered broken for years and is refused here.

  • Genuine content-aware compression requires Ghostscript or MuPDF, both AGPL-3.0 — shipping either to a browser would oblige this entire site to that licence.

  • The widely used open-source PDF library cannot decrypt: its flag for handling encrypted documents lets a broken file through rather than decoding it, and merging one produces blank pages with no error.

  • PDF geometry is measured in PostScript points at 72 to the inch, which makes an A4 page 595 × 842 points.

What we deliberately do not build

Each is something the big PDF services offer. Better to say so here than after you have chosen a file.

PDF to Word, Excel or PowerPoint

No free library can do this with usable fidelity in a browser. The open-source route returns a flat run of paragraphs with the tables, columns and images gone, which is worse than useless for the documents people actually convert. The tools that manage it are commercial SDKs costing four figures. Our PDF to text is the honest subset of it.

Word, PowerPoint or Excel to PDF

Needs an Office rendering engine. There is no browser-side implementation, and the server-side ones would mean uploading your document to us.

HTML to PDF

Needs a headless browser on a server. Your own browser already does it: print the page and choose Save as PDF.

AI summarise and translate

Both need a paid model API, which means sending your document to a third party. If we add it, it will be clearly marked as leaving your device — not slipped in beside tools that do not.

Genuine content-aware compression

Real downsampling needs Ghostscript or MuPDF. Both are AGPL-3.0, and shipping their code to your browser would force this entire site to that licence. Compress PDF offers the two methods that are actually available, and says what each costs.

Removing a password you do not have

Nothing can, and nothing should. Unlock PDF removes protection from a document you can already open.

Keep exploring