What "splitting" a PDF actually means
A PDF is not a stack of flat images; it's a tree of objects. The document catalog points to a page tree, and each page node references its own content streams, fonts, and images (this structure is defined in ISO 32000). Pages are self-contained enough that they can be lifted out of one document and dropped into another without touching anything they don't reference. Splitting a PDF means extracting the objects for the pages you chose — with everything those pages point to — into a brand-new document. The pages you didn't select are simply never copied, and your original file is never modified.
Order and duplicates are features
The output follows the order you type, not the order pages appear in the
source. Enter 5, 1-2 and page 5 comes out first, followed by
pages 1 and 2. Repeat a number — 2, 1, 1 — and that page
appears twice in the result. That turns the tool into a page
re-arranger as much as a splitter: you can reorder, drop, and duplicate
pages in a single pass, without a separate step for each.
Why the result is lossless
Copying happens at the object level, not the pixel level: no page is rendered to an image and no stream is recompressed. Fonts stay embedded, text stays selectable and searchable, and vector graphics stay sharp at any zoom. Compare that with tools that "split" a PDF by rendering each page and re-inserting it as an image — the output looks the same at a glance but produces larger files with text that can no longer be selected or searched.
No upload, verifiable
This tool follows the same privacy model as the merge tool on this site: the browser's File API reads your file locally, extraction runs inside the tab using pdf-lib, and the resulting PDF is assembled in memory and handed to your downloads folder. No server ever receives a byte of your document. You can check this two ways — open your browser's network inspector while extracting and you'll see no upload, or switch your connection off entirely and extract anyway. It still works, because the operation never needed the internet.
The honest limitations
Splitting is page-level, not content-level: you can't cut a page in half or pull out a single paragraph — that's editing, not splitting. Password- protected PDFs are rejected outright, since the tool can't read what it can't decrypt; remove the password in your PDF viewer first. Each run produces one output file from one selection, so extracting every page into its own separate file isn't supported — run the tool once per page as a workaround. And because the whole source file sits in your device's memory while extraction runs, memory is the practical ceiling on how large a file you can split.