Short answer: ChatGPT can read most PDFs in October 2026, including scanned ones, because it now analyzes the images in a PDF as well as its text. When it can't, the cause is one of six things: the file sits in a Project or a GPT instead of the chat, it's over the size or token limit, you've hit an upload cap, the Project is full, the PDF is locked or unusual, or you gave ChatGPT a link instead of a file.
Cause 1 - The PDF is in a Project or a GPT, not the chat
This is now the most common surprise. OpenAI's help centre says files saved as Project files or GPT knowledge "are handled differently from direct chat attachments" and that embedded images "may not be included automatically". A scanned PDF is nothing but images, so in a Project it can look blank even though the same file works when you attach it to a chat.
The fix: attach the PDF directly in the conversation, even inside a Project chat. If you need to ask about one chart or a scanned page, upload that page as a PNG or JPEG. Our guide to ChatGPT Project file limits covers the rest of the Project rules.
Cause 2 - The file is over the size or token limit
Each document can be up to 512 MB and 2 million tokens of text, roughly 1.5 million words. A long or image-heavy PDF can also exceed what fits in a single conversation. When that happens the document may be truncated: ChatGPT reads the first part and answers as if that were the whole thing, which is worse than an error because the answer looks complete.
The fix: split the document into sections and work through them one at a time, or remove pages you don't need.
Cause 3 - You've hit an upload cap
The file is fine; your quota isn't. ChatGPT allows 80 uploads every 3 hours, Free users get 3 a day, and shared storage is capped at 25 GB per user and 100 GB per organization across chats, Projects and GPTs. Failed upload attempts can count toward the rate limit, and ChatGPT doesn't show how much of the rolling quota is left. Both limits surface as "upload limit reached".
The fix: if the error shows a retry time, wait until then. Check Settings, then Storage, and delete files from Library you no longer need: deleting a chat doesn't delete a file saved in Library.
Cause 4 - The Project is full
Projects have their own file count: 5 files on Free, 25 on Go and Plus, and 40 on Edu, Pro, Business and Enterprise. A GPT holds up to 20 knowledge files. Past that, new files are refused even when you have upload quota left.
The fix: remove files the Project no longer needs, merge small related documents into one PDF, or attach the extra file to a single chat instead.
Cause 5 - The PDF is locked or unusual
Password-protected and encrypted PDFs can't be parsed until they're unlocked. PDF portfolios, XFA forms and files exported by unusual tools can also fail even though they open fine in a viewer.
The fix: remove the password if you have the right to, or print the document to a fresh PDF from your viewer and upload that copy.
Cause 6 - You gave it a link or a .gdoc, not a file
ChatGPT can't open links behind a login: Google Drive, SharePoint, a company wiki. OpenAI also says you can't upload a .gdoc file directly; export it as PDF or DOCX first.
The fix: download or export the file and upload it, or paste the relevant text. Better still, use a reader that works on the page you already have open, so there's nothing to download and re-upload.
ChatGPT's file upload limits (October 2026)
All from OpenAI's "Uploading files and audio to ChatGPT" help article, updated 23 October 2026. OpenAI adjusts these, so treat the table as a guide.
| Limit | Value |
|---|---|
| Documents and presentations | 512 MB per file |
| Text and document files | 2 million tokens per file |
| Spreadsheets and CSV | About 50 MB, depending on row size |
| Images | 20 MB per image |
| Upload rate | 80 files every 3 hours (Free: 3 per day) |
| Storage | 25 GB per user, 100 GB per organization |
| Files per Project | Free 5, Go and Plus 25, Edu, Pro, Business and Enterprise 40 |
| Files per GPT | 20 |
| Images inside PDFs | Read in direct chat attachments; may be skipped in Project and GPT files |
| Supported documents | DOCX, PDF, TXT (no .gdoc) |
ChatGPT's error messages, decoded
- "Upload limit reached" - the rate cap or the storage cap (Cause 3). If it shows a retry time, that's the rate cap.
- An answer that ignores your scanned pages, only inside a Project - images in Project files weren't included (Cause 1). Attach the PDF to the chat.
- "This file is too large", or an upload that never finishes - over 512 MB or the token limit (Cause 2). Split it.
- A summary that stops halfway through the document - the conversation ran out of room (Cause 2). Ask about one section at a time.
- "There was an error processing your file" - usually a locked or unusual PDF (Cause 5). Re-save it as a fresh PDF.
- "I can't access external links", or a summary of a login page - a link, not a file (Cause 6). Upload it.
The deeper problem: checking the answer
Even when ChatGPT reads every page, it gives you an answer without an easy way to check it against the document. For a casual question that's fine. For a contract clause, a policy, or a number you'll act on, "it sounds right" isn't good enough, and AI summaries can hallucinate details that were never in the source.
A different approach: read it in the browser, with citations
Instead of uploading a file into a chat, you can question the PDF where it already is. Sidenote is an AI reading assistant for Chrome, Edge and Firefox that answers questions about the PDF you have open, without an upload step:
- Files behind a login - Drive, SharePoint, a wiki - are read in place from your signed-in browser session, read-only.
- Long documents are indexed in full, and each question retrieves the relevant passages, so there's no upload cap or token budget to manage.
- Scanned PDFs are run through OCR on paid plans, so an image-only file becomes searchable, citable text.
- Every answer carries a citation. Click it and the PDF scrolls to the exact sentence and highlights it. Claims that can't be matched to a real passage are dropped server-side before you see them.
ChatGPT vs an in-browser PDF reader
| ChatGPT | In-browser reader (Sidenote) | |
|---|---|---|
| Scanned PDFs | Read in direct chat attachments; may be skipped in Projects | OCR'd on paid plans, then cited like any text |
| Files behind a login | Download and re-upload | Read in place, read-only |
| Very long documents | 2M-token cap; can silently truncate | Indexed in full, relevant passages retrieved |
| Upload caps | 80 per 3 hours; 3 per day on Free | No upload step for open files |
| Verify an answer | Check the PDF by hand | Click a citation to the exact passage |
| Best for | Drafting and reasoning around a document | Reading and questioning real documents |