What a browser can and cannot do to a PDF
PDF Studio is built on two libraries that run entirely on your machine. pdf.js, the renderer Firefox uses to display PDFs, parses your document and draws the page previews. pdf-lib writes the new file. Between them they cover everything on this page, and neither of them has any idea what a server is.
Page-level work — merging, splitting, reordering, rotating, deleting, duplicating, extracting — is exact. The original page objects are copied into the new document untouched, so the text stays selectable, the fonts stay embedded, the links keep working and the file stays the same quality it was. There is no re-encoding step to lose anything in.
Compression is a trade, and we say so
Compressing is different in kind. It renders each page to an image at a lower resolution and rebuilds the document from those images. That is why a 40 MB scan can come out at 3 MB. It is also why the text in the result is no longer selectable or searchable — it is a picture of text now. For a scanned contract you only need to email, that is a good deal. For a document you need to search later, it is not, and the interface says so before you click rather than after.
Where the ceiling actually is
The limit is your device's memory rather than a policy. A 300-page document merges in a couple of seconds on a laptop; the same job on a five-year-old phone takes longer and uses a lot of RAM. Rendering pages to images at 300 dpi is the heaviest thing here — mobile Safari caps total canvas memory and will quietly hand back blank images if you push past it, so the studio frees each canvas the moment it is done with it and warns you before starting something large.
Encrypted files are the one hard stop. If a PDF needs a password to open, it has to be unlocked first — that is what Unlock PDF is for, and it only works when you know the password. Anything else would not be a tool, it would be an attack.