Workbranch › Guides

How to upload files and PDFs to ChatGPT

ChatGPT can read the documents, spreadsheets and images you attach, but each plan has its own limits and long PDFs need a little care. Here's how to attach files, which formats work and how to get better answers from big documents.

Updated 18 October 2026

These steps were checked against OpenAI's help pages on 18 October 2026. Button names may differ slightly between the web, desktop and mobile apps.

Step by step: attach a file

1. Open a new or existing chat on chatgpt.com or in the app.
2. Select the + icon in the message box and choose Add photos & files. On a computer you can also drag the file into the message box, or paste a copied image.
3. Wait until the file has finished uploading.
4. Type what you want: a summary, a comparison of two files, the figures from a table, a rewrite in plainer language.
5. Read the answer and keep asking follow-up questions about the same file in that chat.

Files stored in Google Drive don't have to be downloaded first, if the Google Drive app is available on your plan and in your workspace. Connect the Google Drive app from Plugins or Apps (the label depends on your version), then paste a file link and ask ChatGPT to read it.

Which files work

A Google Docs .gdoc file can't be uploaded directly. Export it as PDF or DOCX first. Video files can be attached as well, including on Free (availability varies by app and account), but OpenAI warns the analysis may skip parts of the video or misread its audio. Videos count toward your upload limits.

Upload limits by plan

• Free: 3 file uploads a day. Documents, spreadsheets, slides and images only; no audio.
• Everyone else: up to 80 files every 3 hours. Paid plans and workspaces can also upload audio.
• Everyone: uploads share a storage cap of 25 GB per user and 100 GB per organization. OpenAI may lower the rate limits at busy times.

Files you add to a project count separately: 5 per project on Free, 25 on Go and Plus, and 40 on Pro, Business, Enterprise and Edu. Our ChatGPT Projects guide covers when that's the better place for documents you reuse.

Tips for long PDFs

• Make sure the text is real text. Outside Enterprise, ChatGPT reads the digital text in a PDF, not the pictures in it. A scanned PDF with no text layer gives it little to work with, so run it through OCR first or upload the pages as images.
• Describe important charts. Images embedded in documents and slides aren't supported yet, so a key chart inside a report may be invisible. Upload the chart separately as an image, or type out its numbers.
• Ask narrow questions. "What does section 4 say about notice periods?" gets a better answer than "tell me about this contract". Name the chapter, heading or page range you care about.
• Work in passes. Ask for an outline of the headings first, then go section by section. Ask it to quote the passage it relied on so you can check it against the original.
• Split very large files. If a document is near the 2 million token cap, or has hundreds of pages you don't need, upload only the relevant part.
• Use spreadsheets for numbers. Tables trapped in a PDF are harder to analyze than the same data as CSV or XLSX, where ChatGPT can count rows and draw charts.

Good to know

• Where files live. Uploads can be saved in Library. To remove one, open Library, select the file and choose Delete. Deleting a file and deleting a chat are separate actions, and text from the file can stay in the chat.
• Training. On personal plans, whether uploads may help improve models depends on your data controls. OpenAI doesn't use content from business offerings such as ChatGPT Enterprise for training.
• Enterprise extra. ChatGPT Enterprise can use visual retrieval for PDFs, which looks at the images too.
• Transcripts aren't perfect. For audio, check names, numbers and who said what against the recording.

If it doesn't work

• "Upload limit reached" after only a few files. Check you're in the right account and plan, then check both the rate limit and your storage. Failed attempts can count toward the limit. Also look at the OpenAI status page.
• The file type is refused. Convert it to one of the formats above, for example a Google Doc to PDF.
• Audio won't upload. The file needs a real audio stream; a WebM or MP4 that is actually a video won't work as audio.
• ChatGPT misses content you can see. It's probably in an image or scanned page. Use OCR or upload that page as an image.

Sources

• Uploading files and audio to ChatGPT (OpenAI Help Center, checked 18 October 2026)
• ChatGPT Image Inputs FAQ (OpenAI Help Center, checked 18 October 2026)
• Are you able to handle images embedded in docs/presentations? (OpenAI Help Center, checked 18 October 2026)
• Google Drive app and setup in ChatGPT (OpenAI Help Center, checked 18 October 2026)

Frequently asked questions