ChatGPT File Upload Limits: How to Handle Large Files and Folders

ChatGPT sets strict limits on file uploads to manage processing load and storage. The hard limit is 512MB per file for documents and presentations, with lower caps for spreadsheets and images. OpenAI gives one upload rate for paid plans and a lower one for Free, and the number of files a project holds varies by plan. If you go over a limit, ChatGPT shows an error and the upload fails.

Official Limits by Plan (2026)

Note: Text and document files are also capped at 2 million tokens per file, so a text-heavy PDF or Word document can hit that cap while it's still under 512MB. The token cap doesn't apply to spreadsheets, whose size limit depends on the size of each row. Upload storage is also capped at 100 GB per organization. OpenAI says it may lower the upload rate during peak hours, and a GPT can hold up to 20 files as knowledge.

Helpful references: Fastio Workspaces, Fastio Collaboration, and Fastio AI.

• ChatGPT Attachment Limit: File Sizes, Counts, and Large Corpus WorkaroundsChatGPT caps message attachments at 20 files, enforces a 512 MB file ceiling, and silently truncates documents beyond 2...
• How to Give ChatGPT Access to Local Files and Synced FoldersConnecting ChatGPT to local files and synced cloud directories eliminates repetitive manual uploads and keeps document...
• ChatGPT Max File Size: The 512 MB Upload Limit Across PlansChatGPT enforces a strict maximum file size limit of 512 MB per uploaded file across all subscription plans, alongside...
• ChatGPT PDF Limit: File Size, Page Count, and Large Document WorkaroundsThe ChatGPT PDF limit enforces a 512MB file size ceiling, a 2 million token extraction threshold, and a restriction of...
• GitHub File Size Limits Explained for Teams and CopilotGitHub hard-blocks any single file larger than 100 MiB in normal Git history, warns above 50 MiB, and caps browser...
• Can ChatGPT Upload Files to Google Drive? How to Save AI OutputsChatGPT cannot natively upload, write, or export files directly back to Google Drive; its native integration is...

More on this subject: Agent File and Document Workflows (325 guides)

OpenAI's list of supported upload types names individual file formats (documents, spreadsheets, presentations, images and audio), not folders or ZIP archives.

The Structure Problem

LLMs work with text streams and individual file contexts. They don't have a file system explorer to navigate directory hierarchies. To work with folders, you need to flatten the structure into a format the model can read, or give the model a tool that can browse folders for it.

Best Workarounds for Folders

1. Upload the Files Individually: Attach the files you need as separate uploads. Each one counts toward the upload rate (80 files every 3 hours on paid plans), so pick the files that matter.
2. Recursive Merging: Use a script to combine all text files in a folder into a single "corpus.txt" file. This works well for codebases but you lose file separation, and the merged file still has to stay under the 2 million token cap.
3. Keep the Folder in a Workspace: Store the folder in a Fastio workspace and install the Fastio plugin in ChatGPT. ChatGPT can then browse the workspace's folders and files and read a file's text, without you uploading each file to the chat.

When your data exceeds 512MB, you need to preprocess it before uploading. Here are three ways to get around the limit.

1. Split and Segment

Break large datasets (like CSVs or logs) into chunks smaller than the limit, which for spreadsheets and CSV files is about 50MB. You can then upload them one at a time. The context window (how much information the AI "remembers" at once) limits how much cross-file analysis you can do.

2. Convert to Text-Efficient Formats

PDFs carry a lot of overhead. Converting data-heavy PDFs to plain text (.txt) can reduce file size without losing the information the AI needs.

3. Use External Link References

Instead of uploading the file directly, host it on a public URL. While ChatGPT can't browse authenticated pages, it can read public text-based content when you point it to a specific link. This bypasses the upload buffer entirely.

Files you upload to ChatGPT stay inside ChatGPT, in the conversation or in your Library, and they count toward its upload storage cap of 25 GB per user. When the same files need to work across sessions, tools and other agents, you need a "file drawer" that lives outside the chat. Fastio provides a dedicated file system for AI agents with generous storage accessible via API. Instead of uploading files to a chat window, your agent connects to Fastio through the Model Context Protocol (MCP) to read, write, and search files instantly.

In ChatGPT itself, the Fastio plugin makes that connection for you. It lets ChatGPT search a workspace by keyword or by meaning, quote the exact passage from a document, and organize files, folders and shares, using your own Fastio account. It doesn't change ChatGPT's own upload limits; the files simply stay in your workspace.

Why Agents Prefer API Storage

• Zero Upload Latency: Agents access files directly from the cloud without uploading them to the chat context window.
• A consolidated MCP toolset: The Fastio MCP server lets agents perform file and search operations that aren't possible in the standard ChatGPT interface.
• Persistent Memory: Files stay in the workspace across sessions, so you can build long-term memory for your AI assistants.
• Query Instead of Paste: With "Intelligence Mode," agents can query large document sets using RAG (Retrieval-Augmented Generation) instead of pasting them into the context window. This changes the model from "chatting with a file" to "giving your AI a hard drive."

Give ChatGPT Your Workspace Files

Install the Fastio plugin to search, read and organize your workspace files from ChatGPT and Codex.

Start 14-Day Trial

In ChatGPT, use the Fastio plugin. It installs from the ChatGPT plugin directory and needs no developer mode:

1. Find Fastio under Plugins in the ChatGPT sidebar.
2. Select Install plugin.
3. Select Connect, then sign in to Fastio in the window that opens and approve access.
4. Type @Fastio in a chat and ask about your files.

The ChatGPT and Codex setup guide also covers installing the plugin in Codex and connecting a custom MCP server instead.

For an agent you run yourself, connect it to persistent storage with the Model Context Protocol.

Step 1: Create a Fastio Workspace Create an organization and start its 14-day trial, which requires a credit card.

Step 2: Connect the Fastio MCP Server Configure your MCP client to connect to Fastio's remote Streamable HTTP endpoint, sending an API key as a Bearer header from an agent that runs without a browser. No local Fastio package or process is required:

{
  "mcpServers": {
    "fastio": {
      "url": "https://mcp.fast.io/mcp/tools",
      "headers": {
        "Authorization": "Bearer your_api_key_here"
      }
    }
  }
}

Step 3: Enable Intelligence Mode In your Fastio dashboard, toggle "Intelligence Mode" on your workspace. This automatically indexes every file you upload (PDFs, code, spreadsheets), making them searchable by meaning, not just keywords. Your agent can now answer questions like "Analyze the financial reports from Q3" without uploading a single file to the chat window.