ChatGPT file upload limits apply in several layers: per-file size, text tokens, upload rate, shared storage and files per project. Passing one check does not mean the upload passes all the others.
OpenAI’s file upload help was updated in October 2026 and checked for this guide on 9 October.
On this page
Size limits by file type
OpenAI lists these current caps:
- documents and presentations: 512 MB per file;
- text and document files: 2 million tokens per file;
- spreadsheets and CSV files: about 50 MB, depending on row size;
- images: 20 MB per image;
- audio files: 512 MB.
The 2-million-token limit does not apply to spreadsheets. A 400 MB PDF can still fail if its extracted text exceeds the token cap.
Upload rate
OpenAI says users can upload up to 80 files every three hours. Free users are limited to three file uploads per day. Limits may be lowered during peak periods.
Failed upload attempts can sometimes count toward the rate limit. ChatGPT does not currently show how much of the rolling upload quota remains.
Storage caps
Shared upload storage is capped at 25 GB per user and 100 GB per organisation. This is different from the rolling rate limit.
Open Settings → Storage to check Library usage when the account has Library. Deleting a chat does not necessarily delete a file saved separately in Library.
Files per project
Project limits depend on plan:
| Plan group | Files per project |
|---|---|
| Free | 5 |
| Go and Plus | 25 |
| Edu, Pro, Business and Enterprise | 40 |
A custom GPT can contain up to 10 knowledge files for the lifetime of that GPT, according to the same help page. Per-file and account storage limits still apply.
Why an upload can fail
Check the limits in this order:
- Confirm the signed-in account and plan.
- Check the file extension and whether ChatGPT supports it.
- Check file size.
- For documents, consider the extracted token count.
- Check recent upload attempts and the rolling rate.
- Check Library or workspace storage.
- Check OpenAI Status for an incident.
Google Docs files cannot be uploaded directly. Export a .gdoc as PDF or DOCX first.
Images inside PDFs
OpenAI says ChatGPT Enterprise supports visual retrieval for PDFs. Other plans and document types may use text-based retrieval, which extracts digital text without reading embedded images in the same way.
If a chart or scanned page matters, verify that the plan and file type support it. Otherwise upload the image separately or provide OCR text.
Retention is separate from limits
Deleting a saved Library file and deleting the conversation are separate actions. Files in projects, GPT knowledge and workspaces can follow different retention rules.
Do not use upload limits as a privacy assumption. Read the service’s data controls and the policy for the account type.
