Why can't ChatGPT read my PDF?

Start with the diagnosis, because the error message doesn't tell you which problem you have. Open the PDF on your computer and try to select a line of text with your mouse. That one gesture separates the most common cause from all the others.

Cause 1: your PDF is a scan — there's no text inside

If you can't select text — the page is one solid image — your PDF is a scan or a photo wrapped in a PDF. There are no words in the file, only pictures of words. When ChatGPT says “no text could be extracted from this file”, this is almost always why: its text extraction found nothing to extract, because nothing is there.

The fix is character recognition (OCR), and you have two easy routes:

• Export the PDF's pages as images (most PDF viewers offer “export as PNG/JPG”), then run them through an OCR tool — our image to Markdown converter does this in your browser, without the scan leaving your device.
• If the document matters long-term, OCR it once properly and save the result — a scan you query often is worth converting once, not re-processing every chat.

Cause 2: the PDF is password-protected

An encrypted PDF can't be opened by ChatGPT's file processing at all — the content is mathematically sealed. The symptom is usually a flat failure rather than a partial answer. The fix is to remove the protection yourself first: open the PDF with its password, print-to-PDF or “save as” an unprotected copy, and upload that. No online service can (or should) do this without the password.

Cause 3: the file is too big — in megabytes or in tokens

ChatGPT's documented upload limits are generous on paper — 512 MB per file, with text files capped at 2 million tokens — but a long PDF can also simply exceed what the model will attend to well in one go. The practical symptoms: the upload succeeds but answers only cover the first part of the document, or the file is rejected outright.

The fix that actually works is shrinking the content, not the file. Compressing a PDF reduces megabytes but not tokens — the words are the words. Converting the PDF to Markdown reduces both: the formatting machinery disappears, and the same content arrives as lean text. A document that fails as a PDF very often fits comfortably as a .md file.

Cause 4: you've hit an upload cap, not a file problem

Free ChatGPT accounts have a small daily file allowance, and paid plans meter uploads in windows (80 files per 3 hours, per OpenAI's FAQ). If the error appeared after several uploads — especially “you've reached our limit of file uploads” — the PDF is fine and the meter is the problem. Waiting works. So does needing fewer uploads: paste converted Markdown into the chat instead of attaching files, and the file meter never ticks.

The fix that covers all four at once

Notice the pattern: every cause is about the container, never the content. The words in your document are fine — the PDF wrapper is what's scanned, sealed, oversized or metered.

So remove the wrapper. Convert the PDF to Markdown once, in your browser, and paste the result. Pasted text can't be “unreadable”, has no upload meter, and costs a fraction of the tokens — AI chats process PDF pages as images as well as text, which per Anthropic's documentation multiplies the cost roughly sevenfold versus plain text. Our free PDF to Markdown converter keeps headings, paragraphs and tables, shows you the token count before and after, and never uploads your file anywhere.