# ChatGPT File Limit: How Many Files You Can Upload and Storage Caps ChatGPT restricts file uploads to 10 files per message turn, 512MB per document or audio file, and 2 million tokens per text document. Users also face account storage quotas of 25 GB per individual and 100 GB per organization. Connecting an external indexed workspace through the Model Context Protocol allows teams to reference large document corpora without hitting chat attachment limits. Source: https://fast.io/resources/chatgpt-file-limit/ Author: [Derek Labian](https://fast.io/authors/derek-labian/) Last reviewed: 2026-10-31 ## What Are the Current ChatGPT File Upload Limits? According to [OpenAI Help Center documentation](https://help.openai.com/en/articles/8555545-uploading-files-and-audio-to-chatgpt) updated in October 2026, ChatGPT restricts uploads to 10 files per message turn, 512MB per file, 2 million tokens per document, and 80 files per rolling 3-hour window, with total shared upload storage capped at 25 GB per user and 100 GB per organization. The ChatGPT file limit restricts users to uploading up to 10 files per conversational message turn, with individual file sizes capped at 512MB and total lifetime account storage capped by plan tier. When you drag and drop files into the ChatGPT prompt box, you encounter three distinct enforcement layers: * **Per-turn attachment limits:** You can attach a maximum of 10 files in a single prompt message. If you select 11 files in your system file picker, the ChatGPT interface rejects the batch until you deselect at least one item. * **Per-file size limits:** Documents and presentations uploaded to ChatGPT carry a maximum physical size limit of 512MB per file. Audio uploads on paid ChatGPT plans are also capped at 512MB per file. Spreadsheets and comma-separated value files follow a stricter 50MB ceiling, while standalone images are limited to 20MB per image. * **Token ceilings:** Text-heavy document files uploaded to a ChatGPT conversation cannot exceed 2 million tokens per file. A document that weighs only a few megabytes on disk can still fail to upload if its dense technical text exceeds this token budget. Beyond what fits into a single prompt, OpenAI enforces velocity caps on how rapidly you can process files. Unpaid ChatGPT accounts are limited to 3 file uploads per day. Subscribers on paid tiers (Plus, Pro, Team, and Enterprise) can upload up to 80 files every 3 hours. OpenAI reserves the operational right to reduce these rolling limits during periods of high infrastructure demand. These limits govern the native chat interface across web and mobile applications. They exist to prevent conversation context windows from becoming saturated and to keep server-side sandbox environments from running out of memory during code execution. ## How Do File Limits Compare Across ChatGPT Plans and Claude? File handling rules vary depending on your subscription tier and whether you use transient chat attachments, persistent project files, or custom GPT knowledge stores. The following comparison table outlines documented file restrictions across ChatGPT tiers and compares them against Anthropic's Claude: | Platform / Plan | Files per Turn | Single File Size Limit | Total Storage Quota | Rolling Rate Limit | Date Verified | | :--- | :--- | :--- | :--- | :--- | :--- | | ChatGPT Free | 10 files | 512MB (doc/presentation), 50MB (spreadsheet), 20MB (image) | 25 GB per user quota | 3 uploads per day | October 2026 | | ChatGPT Plus & Pro | 10 files | 512MB (doc/presentation/audio), 50MB (spreadsheet), 20MB (image) | 25 GB per user quota | 80 files per 3 hours | October 2026 | | ChatGPT Team & Enterprise | 10 files | 512MB (doc/presentation/audio), 50MB (spreadsheet), 20MB (image) | 100 GB per organization quota | 80 files per 3 hours | October 2026 | | Claude Pro & Team | 20 files | 500MB (chat), 30MB (project) | Bounded by context window | Dynamic rate limit | September 2026 | ### Project Files Versus Direct Chat Attachments In ChatGPT, project spaces allow teams to organize conversations around shared documentation. The number of files you can attach to a project workspace depends strictly on your plan level: * **ChatGPT Free accounts:** Up to `5` files per project. * **Plus and Pro tiers:** Up to `25` files per project. * **Team, Enterprise, and Edu tiers:** Up to `40` files per project. Custom GPTs follow a different ceiling. When building a custom GPT, you can upload up to `20` files as foundational knowledge. These files count toward your account storage quota and remain accessible across all sessions directed to that custom GPT. In contrast, [Claude documentation](https://support.claude.com/en/articles/8241126-upload-files-to-claude) confirms that Claude accepts up to 20 files per individual chat, with each file supporting up to 500MB. Inside Claude Projects, files are capped at 30MB each, but the overall project file count is unlimited, provided the extracted text fits within the model context window. Understanding these differences helps teams select the right structure before uploading large document sets. ## Why Does ChatGPT Reject Large Documents and Spreadsheets? When ChatGPT fails to process an upload, the error message rarely explains the underlying technical constraint. The system evaluates files across several architectural barriers, each with its own failure trigger. ### The Token Ceiling Versus Physical File Size The 512MB cap applies to raw binary storage, but the 2 million token cap governs the text parser. Tokenization converts text into sub-word units. In standard English prose, a 2 million token allocation accommodates roughly 1,500,000 words across uploaded documents in ChatGPT, corresponding to thousands of pages. However, formatted data and source code tokenize with much lower efficiency: 1. **Uncompressed text files:** A plain text file containing dense JSON records, SQL dumps, or system logs can contain millions of tokens while remaining under twenty megabytes on disk. When uploaded, ChatGPT rejects the file with an error stating the document exceeds limits, even though its file size is far below 512MB. 2. **Dense academic and legal PDFs:** Documents filled with tables, formulas, citations, and dense formatting produce high token densities. When the text extraction pipeline hits the 2 million token limit for text files uploaded to ChatGPT, extraction halts or fails entirely. ### Spreadsheet Complexity and Sandbox Memory Under OpenAI guidelines, spreadsheets and CSV files operate under an explicit 50MB ceiling, but their real limit is memory consumption during execution. When you upload a tabular dataset, ChatGPT loads the file into a Python execution sandbox (formerly Advanced Data Analysis) using libraries such as Pandas. If a CSV file contains hundreds of columns or millions of rows with wide string fields, loading the dataset into a Pandas DataFrame can expand memory usage by multiple times the original file size. A spreadsheet file uploaded to a ChatGPT conversation approaching the 50MB limit can expand into substantial memory allocations during ingestion. If memory allocation exceeds sandbox container limits, the Python kernel terminates abruptly, returning an uninformative execution error. ### PDF Page Parsing and Visual Elements PDF processing changes behavior based on document length: * **Documents under 100 pages:** ChatGPT analyzes both the raw text stream and visual elements, including embedded images, diagrams, charts, and mathematical notation. * **Documents from 101 to 1,000 pages:** ChatGPT processes text streams only. Embedded charts, illustrations, and scanned diagram content are omitted from analysis. * **Documents exceeding 1,000 pages:** The upload pipeline rejects the file outright. ### Audio Format Restrictions Audio uploads are restricted to paid subscription workspaces. ChatGPT accepts audio files up to a 512MB maximum size limit across standard formats including WAV, MP3, OGG, AAC, and M4A. However, container formats such as WebM and MP4 must contain an audio-only stream. If the ingestion pipeline detects a video track inside a WebM or MP4 container, the upload fails immediately. ## How Can You Manage and Clear Stored Files in ChatGPT? Every file uploaded to a ChatGPT conversation, project, or custom GPT is automatically cataloged in your account's persistent storage repository, known as the Library or Space. Over time, these files accumulate toward your account quota. Many users assume that deleting a conversation thread removes the associated files from OpenAI servers. This assumption is incorrect. Deleting a chat removes the dialogue history from your sidebar, but the underlying uploaded files remain stored in your Library, continuing to draw down your 25 GB user quota or your organization's 100 GB quota. ### Auditing Account Storage To inspect your current storage utilization on the web interface: 1. Open your account profile menu in the bottom-left corner of the ChatGPT window. 2. Select the settings menu option to open account preferences. 3. Choose the storage section to view your total consumed gigabytes and inspect individual stored assets. The storage dashboard displays your current total against the 25 GB user account quota ceiling. However, ChatGPT does not display a counter for your rolling 80-file upload velocity quota. If you hit the rate limit, you must wait for the rolling 3-hour window to advance. ### Deleting Files to Reclaim Quota To free up storage space permanently: 1. Open your Library or Space view from the main interface. 2. Select the specific document, dataset, or audio recording you wish to remove. 3. Select the delete action to move the file to your recently deleted folder. 4. If your interface provides an option to clear immediately, choose the permanent deletion command. Once permanently deleted, the allocated capacity returns to your available account balance. 5. For custom GPTs, open the GPT builder, locate the knowledge section, and delete outdated reference files manually. In organization environments running Team or Enterprise subscriptions, workspace administrators can configure data retention schedules. These policies automatically purge files after designated intervals, preventing accumulated uploads from breaching the shared 100 GB organization ceiling. ## How to Bypass ChatGPT File Limits Using Fastio and MCP Attaching files directly to chat prompts is an inefficient way to work with large document libraries. Every file you attach consumes valuable context window space, risks hitting upload rate limits, and remains isolated within a single conversation thread. When you start a new conversation, you must upload the files again. Before considering external indexing solutions, teams often attempt several manual workarounds: * **Splitting files manually:** Splitting a large PDF or voluminous dataset into smaller sections allows each piece to stay under token limits. However, this process destroys cross-document context, making it impossible for the model to synthesize facts across sections. * **Using cloud storage links:** Pasting public sharing links from Google Drive, Dropbox, Box, or OneDrive allows ChatGPT's web browsing tool to read individual pages. However, the browsing tool cannot search across thousands of files, read nested folder hierarchies, or execute complex semantic queries. * **Building custom vector databases:** Engineering teams sometimes deploy custom Python retrieval-augmented generation pipelines connected to cloud object storage. While functional, this approach requires provisioning infrastructure, maintaining vector indices, and writing custom authentication code. ### The Intelligent Workspace Alternative A more effective approach connects ChatGPT directly to an external, intelligent workspace using the Model Context Protocol (MCP). Instead of pushing files into the chat interface, your files remain stored and indexed in a dedicated workspace, and ChatGPT queries that workspace on demand. [Fastio workspaces](/product/workspaces/) provide persistent environments designed for agentic teams. In an [intelligent workspace](/product/ai/), files are indexed automatically for hybrid search (combining full-text keyword matching, semantic search, and metadata filtering). For structured document processing and attribute tracking across large corpora, [Metadata Views](/product/document-data-extraction/) extract typed database schemas directly from files without manual rules. When ChatGPT needs information, it searches the workspace and retrieves only the exact passages required to answer your prompt, accompanied by citations. This architecture changes the operational dynamic: * **No attachment limits:** Your workspace can hold thousands of documents, high-resolution media, and massive datasets. Fastio supports single-file uploads of `25 GB` on Starter, `50 GB` on Business, and `100 GB` on Enterprise, far exceeding standard AI chat upload boundaries. * **Zero context waste:** Rather than stuffing millions of tokens of raw document text into the conversation, the model retrieves only the relevant paragraphs. * **Multi-cloud synchronization:** Cloud Sync connects Google Drive, Dropbox, Box, and OneDrive folders to a workspace, one-way or two-way, on a schedule or on demand. Existing company folders become searchable without manual migration. * **Version preservation:** Every document retains its complete per-file version history, allowing team members and assistants to track changes over time. * **Detailed audit visibility:** A detailed activity log records changes to workspace assets, including uploads, edits, moves, and deletions, without recording read-only access events. ### Step-by-Step: Connecting ChatGPT to Fastio To connect ChatGPT to a Fastio workspace, you can use the plugin directory or configure a custom remote MCP server: #### Method 1: The Fastio Plugin (Recommended) 1. Open the ChatGPT plugin directory by visiting `chatgpt.com/plugins`. 2. Search for **Fastio** in the directory listings. 3. Select **Install plugin** and complete the standard sign-in prompt to authorize access to your Fastio organization. 4. In any ChatGPT conversation, invoke the connection by mentioning `@Fastio` alongside your query. ChatGPT will search your indexed workspace files to answer questions. #### Method 2: Custom MCP Server via Streamable HTTP If your workspace policies require direct endpoint configuration: 1. Open ChatGPT settings, access the security section, and turn on Developer mode if enabled by your organization. 2. In the plugin settings, select the option to add a custom server. 3. Name the connector `Fastio`. 4. Enter `https://mcp.fast.io/mcp/operations` as the remote MCP server URL. The server operates over Streamable HTTP. Complete setup steps and technical documentation are available in the [Fastio MCP setup guide](https://mcp.fast.io/docs), the [agent guide](https://mcp.fast.io/skill.md), and [storage for agents](/storage-for-agents/). 5. Complete the sign-in authorization in the browser window that appears. You can review and approve workspace permissions before finalizing the connection. Once connected, ChatGPT queries your documents through actions exposed by the server. You can ask: *"What are the payment terms outlined in the master services agreements in my legal workspace?"* ChatGPT queries the indexed files, identifies the relevant contracts, and returns the answer with precise document citations, completely bypassing the native 10-file attachment cap. Fastio monthly plans start with a 14-day free trial, which requires a credit card. Paid subscriptions include Starter at `$9.99/mo` (3 seats, `250 GB` storage, 5 workspaces, `100,000` credits a month), Business at `$49.99/mo` (10 seats, `5 TB` storage, 50 workspaces, `600,000` credits a month), and Enterprise at `$199.99/mo` (30 seats included, additional seats at `$1` a month each up to `200`, `25 TB` storage, 200 workspaces, `3,000,000` credits a month). Details are published on the [Fastio pricing page](/pricing/). Annual plan pricing is `$99`, `$499`, and `$1,999` a year. ## Frequently asked questions ### How many files can I upload to ChatGPT at once? You can attach up to 10 files in a single prompt message uploaded to a ChatGPT conversation. If you select more than 10 files in your file chooser, the interface will prompt you to remove files before sending the message. ### What is the total file limit for ChatGPT Plus? ChatGPT Plus allows users to upload files up to 512MB each, with a rolling rate limit of 80 files every 3 hours. Total lifetime storage is capped at 25 GB per user account, and each project space can hold up to 25 files. ### Does ChatGPT keep uploaded files permanently? Uploaded files remain in your account Library or Space until you delete them manually or until an organization retention policy purges them. Deleting a chat thread does not delete the files stored in your Library. ### Why does ChatGPT say my file is too large when it is under 512MB? Text files and PDFs are bound by a 2 million token limit in addition to the 512MB physical limit. Dense text files, logs, or large documents can easily exceed 2 million tokens while remaining under twenty megabytes in file size. Spreadsheets uploaded to a ChatGPT conversation also carry a strict 50MB file size limit. ### What happens when I hit the 80 files per 3 hours limit? When you exceed the rolling rate limit, ChatGPT displays an upload limit reached message and prevents further uploads until the 3-hour window advances. Failed upload attempts can sometimes count against this quota. ### Can ChatGPT read Google Docs or spreadsheets directly? ChatGPT cannot read Google Docs files (.gdoc) directly; they must first be exported to PDF or DOCX format. Spreadsheets in XLSX, XLS, CSV, or TSV formats can be uploaded directly up to approximately 50MB. ## Sources - [OpenAI Help Center: Uploading files and audio to ChatGPT](https://help.openai.com/en/articles/8555545-uploading-files-and-audio-to-chatgpt): ChatGPT accounts allow users to upload up to 10 files per prompt message, with unpaid accounts limited to 3 file uploads per day and paid accounts allowed up to 80 files every rolling 3-hour window with shared account storage capped at 25 GB per user and 100 GB per organization, while individual documents have a maximum size limit of 512MB per file and text files are capped at 2 million tokens per file. - [OpenAI Help Center: Uploading files and audio to ChatGPT](https://help.openai.com/en/articles/8555545-uploading-files-and-audio-to-chatgpt): Text and document files uploaded to a ChatGPT conversation, project workspace, or custom GPT are capped at up to 2 million tokens per file with spreadsheets limited to approximately 50MB and images limited to 20MB. ## About Fast.io Fast.io provides shared workspaces where people and AI agents work on the same files, with built-in semantic search and citation-backed chat over what they hold. Agents reach it through a remote MCP server, a REST API at https://api.fast.io/current/, and a command line client published on npm as @vividengine/fastio-cli. MCP setup is at https://mcp.fast.io/docs: Claude and most MCP clients connect to https://mcp.fast.io/mcp/tools, ChatGPT to https://mcp.fast.io/mcp/operations, and coding agents to https://mcp.fast.io/mcp/code.