Listen to a PDF

No. The file is opened and read inside this tab, from your computer’s memory — there is no upload and no copy on any server. That is also why it works on a document you would never hand to a website: a contract, a medical letter, an unpublished manuscript.

Because it is a scan. A PDF made by a scanner or a phone camera is a stack of photographs, and there are no words in it to read — the text you see is part of the picture. Getting text out of one needs OCR, which no web page can do quickly. If you can select a sentence with your cursor in another PDF viewer, this page can read it; if you cannot, nothing here can.

Yes, from the premium AI voice mode — a real MP3 file to keep, with no watermark and no length paywall. The free browser voice cannot be downloaded, because browsers do not let a web page record their built-in speech engine to a file. That is a limit of the browser, not a restriction we added.

Two reasons, and neither is a paywall trick. What is on screen is what the highlighting can follow; and on the AI voice you are paying your own provider by the character, so a 400-page report in one click would be a genuinely unpleasant surprise. Reading whole documents end to end, remembering where you stopped and syncing that across devices is what the full AllReads app is for.

Because this reads the PDF rather than drawing it. The words are pulled out and set in one clean reading column, so the type size is yours to choose, the line length stays readable on a phone, and the highlight can follow real words. Columns, headers and footers are unwound in the process — a page number in the margin is dropped rather than read out mid-sentence.

No. The key is sent straight from your browser to OpenAI or xAI and nowhere else — it never reaches an AllReads server, and it is not written to your browser’s storage either. Closing the tab discards it. You can confirm this in your browser’s network tab: the only request carrying the key goes to the provider’s own domain.

Files up to 100MB. The document is parsed in your browser’s memory and each page’s text is pulled out as you reach it, so page count matters far less than file size — a 600-page book of text opens faster than a 40-page brochure full of photographs.