ChatGPT File Upload Limit

ChatGPT file upload limits
in 2026

ChatGPT has a 512 MB hard limit per uploaded file. Text and document files also have a 2 million-token cap, so a document can be under 512 MB and still exceed the limit. Spreadsheets, images, upload frequency and total account storage have separate caps.

Last verified September 3, 2026 · Limits may vary by plan and can be reduced during peak hours

Quick Answer

What is the ChatGPT file upload limit?

Limit typeOpenAI-published limit
Any single file512 MB per file
Text or document file2 million tokens per file
CSV or spreadsheetApproximately 50 MB, depending on row size
Image20 MB per image
Rolling upload rateUp to 80 files every 3 hours
Free plan upload rate3 file uploads per day
End-user storage25 GB
Organization storage100 GB
Files attached as custom GPT knowledgeUp to 10 files over the lifetime of that GPT
Files in a ChatGPT ProjectPlus: up to 20; Pro, Team, Education and Business: up to 40

These limits apply in different ways. A 300 MB PDF may be below the file-size cap but above the 2 million-token document cap. A small file may also fail if the account has reached a rolling upload-rate or shared storage limit. OpenAI notes that upload limits may be lowered during peak hours.

Source: OpenAI File Uploads FAQ →

Why An Upload Can Fail Below 512 MB

“File too large” is not the only limit

The document exceeds 2 million tokens

Text documents have both a byte-size limit and a token limit. Whichever limit is reached first controls.

The spreadsheet exceeds its practical cap

CSV and spreadsheet uploads are limited to approximately 50 MB, with the exact threshold affected by row size.

The image exceeds 20 MB

Images have a separate per-image limit rather than the general 512 MB ceiling.

The rolling rate limit has been reached

Paid users can generally upload up to 80 files every three hours. Free users are limited to three uploads per day. Failed attempts can count toward the cap.

Shared storage is full

The storage cap is shared across chats, Projects and custom GPT knowledge: 25 GB per end user and 100 GB per organization.

The service is experiencing an incident

If the file and account appear within the limits, check OpenAI's service-status page before repeatedly retrying.

Troubleshooting

What to do when ChatGPT says “upload limit reached”

1

Identify the type of limit

Check the file's byte size, likely token count, file type, recent upload activity and account storage. These are separate limits.

2

Confirm the account and plan

Make sure you are signed in to the intended account and workspace. Plan-specific controls can change the available quota.

3

Check shared storage

Review Library storage under Settings → Storage. Deleting unneeded chats or custom GPT knowledge can free shared file storage.

4

Avoid repeated failed attempts

Failed uploads may count toward the rolling rate limit. Diagnose the likely limit before retrying.

5

Check service status

If the limits do not explain the error, review OpenAI's status page for an active file-upload incident.

The Bigger Question

Uploading the file is not the same as analyzing the complete record

A document may fit under every upload limit and still be difficult to investigate as part of a large file set.

The relevant clause may be in one contract.

The later amendment may change it.

A spreadsheet may conflict with a number in the narrative.

An exhibit may contradict what a witness said.

The question is not

Can I upload this file?

It is

Can the system analyze the complete document set together and show me the evidence behind every material finding?

Continua is built for long, multi-file professional investigations with up to a 1 billion-token context window, complete-record analysis and cited, reviewable outputs.

Ask One Question

Ask the complete file set one question

Analyze this complete document set together. Identify the key events, obligations, figures, changes and unresolved issues. Reconcile facts across the documents, flag contradictions and missing support, separate confirmed findings from open questions, and cite every material finding to the exact source page.
Try This on Your Files Free · No card required
When Continua Is The Better Fit

Built for investigations that span files

Long legal matters

Analyze transcripts, exhibits, contracts and correspondence as one evidence set.

Medical-record review

Build a chronology across records from multiple providers and flag treatment gaps or conflicting histories.

Due diligence

Reconcile claims in a CIM or presentation against financials, tax returns, leases and contracts.

Contract analysis

Determine the effective terms after amendments, schedules, addenda and later agreements are considered.

RFP and bid compliance

Compare requirements across the solicitation, amendments, questions, answers and proposal package.

Structured-data investigations

Analyze documents and spreadsheets together rather than treating each file as an isolated chat attachment.

How It Works

From file pile to cited report

1

Attach the complete record set

Add the documents and datasets that belong to the investigation.

2

Describe what must be determined

State the decision, disputed issues, comparisons, thresholds and exceptions that matter.

3

Set the criteria and guardrails

Define what counts as a finding and how the final report should be organized.

4

Review the report and evidence

Inspect cross-document findings and move from each material claim to the supporting source.

5

Follow up or export

Ask additional questions, create a chronology or table, and export the completed report.

Scope

A long-document investigation workspace — not an upload-limit bypass

Continua does not claim unlimited uploads, and it does not change ChatGPT's limits. It is a separate workspace designed for professional analysis of large, multi-file records.

Use ChatGPT for general conversational work and ad hoc analysis. Use Continua when the task depends on complete-record reasoning, cross-document reconciliation and inspectable source citations.

Comparison

ChatGPT file analysis vs. Continua investigations

ChatGPTContinua
General-purpose conversational AIInvestigation workspace for professional document sets
512 MB hard limit per fileProduct advantage is complete-record analysis, not an “unlimited upload” claim
2 million tokens per text or document fileUp to a 1 billion-token context window
Separate rate and storage capsDesigned to analyze long, multi-file records together
Useful for questions about individual filesBuilt to reconcile facts, changes and conflicts across files
Output may require manual source checkingMaterial findings remain linked to supporting evidence

Actual product limits and availability can change. Verify current limits before relying on this comparison.

Useful Questions

Questions to ask before choosing a file-analysis tool

How many files belong to the investigation?
Will the system analyze them together or one at a time?
Is the relevant content text, tables, images, audio or video?
Does the task require exact source citations?
Must later amendments override earlier terms?
Do figures need to be reconciled across documents and spreadsheets?
Should contradictions be surfaced rather than summarized away?
Are temporary files deleted after the task?
Will uploaded data be used to train AI models?
Does the report need to be exported for professional review?

Supported formats include PDF, DOCX, XLSX, CSV and other document, audio and video formats shown in the uploader.

Zero document retention for normal tasks

  • Files attached directly to a normal task are deleted after the task completes
  • Customer documents, prompts and task data are not used to train AI models
  • Files remain persistent only when intentionally added to a Project
  • Reports can be exported and findings remain linked to supporting evidence
FAQ

Frequently asked questions

What is the maximum ChatGPT file upload size?

ChatGPT has a hard limit of 512 MB per file. Text and document files also have a 2 million-token cap, so the token limit may be reached before the byte-size limit.

What is the ChatGPT PDF upload limit?

A PDF is subject to the 512 MB per-file limit and, when treated as a text document, the 2 million-token cap. On non-Enterprise plans, OpenAI says document retrieval is text-based and embedded images are discarded; Enterprise supports visual retrieval for PDFs.

How many files can I upload to ChatGPT?

OpenAI publishes several separate limits: up to 80 files every three hours for users generally, three uploads per day for Free users, up to 10 knowledge files over the lifetime of a custom GPT, and separate per-Project limits by plan. Limits can be lowered during peak hours.

Why does ChatGPT say “upload limit reached” when I uploaded only a few files?

You may have reached a shared storage cap rather than the rolling upload rate. Failed attempts can also count toward the rate limit, and uploads across chats, Projects and custom GPT knowledge share storage caps.

Does splitting a large document solve the limit?

Splitting can bring individual files below a size or token cap, but relevant facts may be distributed across separate uploads or conversations. For important work, confirm that the complete record is analyzed together and that findings remain traceable to their sources.

Is the ChatGPT API file limit the same as the ChatGPT product limit?

Not necessarily. ChatGPT and the OpenAI API are separate products with different upload purposes, storage rules and limits. Check the current documentation for the product and workflow you are using.

Does Continua have unlimited file uploads?

No unlimited-upload claim is made here. Continua's advantage is an investigation workflow built around long, multi-file record sets, with up to a 1 billion-token context window and cited, reviewable findings.

Is Continua part of OpenAI or ChatGPT?

No. Continua is an independent product. ChatGPT is a product of OpenAI. The limits on this page are provided for comparison and troubleshooting.

Keep Exploring

Related investigation skills, side-by-side comparisons and the accuracy benchmarks behind this workflow.

First task free

Run Your First Investigation Free

Attach the file pile, ask the hard question, get back a cited report with every conflict flagged. No credit card required.

No credit card required
Zero document retention
Cited, reviewable outputs