# Claude Upload Limits: Documented File Size, Count, and Context Ceilings Claude upload limits differ between standard chats and Claude Projects. Chats accept up to 20 files at 500MB each, while Projects accept 30MB files with no fixed file count, constrained entirely by whether the text fits within the model's context window. For large document collections that exceed these limits, external workspace indexing with remote search provides full retrieval without consuming context. Source: https://fast.io/resources/claude-upload-limit/ Author: [Tom Langridge](https://fast.io/authors/tom-langridge/) Last reviewed: 2026-11-07 ## What Are the Documented File Upload Limits for Claude? According to Anthropic's documented support specifications, Claude standard chat conversations accept up to 20 files with a maximum file size of 500MB per file. In comparison, Claude Projects accept files up to 30MB each with no fixed file count cap, bounded by context window capacity. Claude upload limits differ by mode: standard chats accept up to 20 files at 500MB each, while Claude Projects permit files up to 30MB each with no fixed file count, constrained entirely by whether total text fits into Claude's context window. Understanding these restrictions requires separating network transport ceilings from active model attention limits. The 500MB file limit in a standard chat is an HTTP transport threshold. It governs the raw payload size that Anthropic's servers accept during upload. The context window is an active memory threshold. It determines how many tokens Claude can evaluate during an ongoing conversation. A single plain text export or an uncompressed dataset can contain millions of tokens, exceeding Claude's `200,000`-token context capacity while remaining well below the 500MB transport boundary. As of November 2026, the following comparison table details documented upload thresholds, file caps, and processing rules across Claude interfaces: | Upload Mode | Max File Size | File Count Limit | Context & Processing Ceilings | Date Checked | Source | | :--- | :--- | :--- | :--- | :--- | :--- | | Standard Chat (claude.ai & Desktop) | 500MB per file | Up to 20 files per chat | Shared `200,000`-token context window; PDFs capped at `1,000` pages | 2026-11-07 | [Anthropic Help Center](https://support.claude.com/en/articles/8241126-upload-files-to-claude) | | Claude Projects (Free Tier) | 30MB per file | Unlimited (fits context) | Standard `200,000`-token context window; text extraction only | 2026-11-07 | [Anthropic Help Center](https://support.claude.com/en/articles/8241126-upload-files-to-claude) | | Claude Projects (Pro & Team Plans) | 30MB per file | Unlimited (fits context) | `200,000`-token active window with automated retrieval expansion | 2026-11-07 | [Anthropic Help Center](https://support.claude.com/en/articles/8241126-upload-files-to-claude) | | Claude Files API (Developers) | 500MB per file | Org storage quota (`1 TB` default) | Referenced by file ID; context tokens charged on prompt inclusion | 2026-11-07 | [Anthropic Help Center](https://support.claude.com/en/articles/8241126-upload-files-to-claude) | Standard chat uploads persist only for the lifetime of an individual conversation thread. When you attach spreadsheets, PDFs, or images to a chat, Claude parses the content directly into the conversation history. In contrast, Claude Projects act as persistent workspaces where attached files remain accessible across multiple conversation sessions within that project. ## How Do Claude Projects Handle Knowledge Bases and Context Ceilings? A widespread misconception across third-party comparison guides is that Claude Projects enforce a strict cap of five, ten, or twenty files. In reality, Anthropic documents no fixed file-count limit for Claude Projects. Claude Projects accept files up to 30MB each with no fixed file count cap, bounded by context window capacity. You can upload dozens of individual files into a project knowledge base, provided each file stays within the 30MB ceiling and the cumulative text fits within the context window. When you add documents to a Claude Project, the system extracts the plain text content and stages it in the project knowledge base. Unlike standard chats, where files must be re-attached for new tasks, project files are automatically available as background reference for every prompt sent within that project workspace. The practical bottleneck for Claude Projects is token density rather than the quantity of files: - **Token Consumption Per Document:** Formatted text files, source code repositories, and dense spreadsheets generate tokens rapidly. A lengthy technical manual or a collection of legal briefs can easily reach `150,000` to `250,000` tokens. - **Conversation Headroom:** In standard project configurations, project knowledge shares the `200,000`-token context window with your system prompt, custom project instructions, ongoing conversation turns, and Claude's responses. Staging `180,000` tokens of project files leaves only `20,000` tokens for your actual dialogue, leading to early conversation truncation. - **Automated Retrieval Expansion:** On paid Pro, Team, and Enterprise accounts, Anthropic deploys automated retrieval-augmented generation for project knowledge that approaches the context ceiling. Rather than stuffing the entire knowledge base into the active prompt on every turn, Claude indexes the text and retrieves relevant passages on demand. - **Text Extraction Boundaries:** Project files support text extraction only, with the exception of multimodal PDFs. If you upload Word documents (.docx) or Markdown files (.md) containing embedded charts, diagrams, or graphics, Claude extracts only the written text and ignores the images completely. When managing extensive document libraries across multiple project collaborators, relying solely on in-browser uploads creates operational friction. Teams working with hundreds of megabytes of documentation need a reliable mechanism to stage files externally and supply targeted passages on demand. ## Why Does Claude Return File Too Large or Context Exhaustion Errors? Users frequently encounter upload errors when interacting with Claude. An error message stating that an uploaded file is too large does not always mean the file crossed the Claude standard chat 500MB file limit. Anthropic triggers this rejection across several distinct failure conditions. The following failure modes explain why Claude rejects file uploads and how to resolve them: ### 1. The 1,000-Page PDF Ceiling Anthropic enforces a strict hard limit of `1,000` pages for any uploaded PDF document. If you upload a PDF containing `1,001` pages, Claude rejects the file immediately, regardless of whether the file size is `5 MB` or `450 MB`. Furthermore, Claude applies two tiers of PDF analysis: - **`100` Pages or Fewer:** Claude processes both text and visual elements, including charts, diagrams, tables, and illustrations. - **`101` to `1,000` Pages:** Claude processes text only. All visual elements, embedded images, and infographics are omitted from model interpretation. If you must analyze documents exceeding `1,000` pages, split the document into separate volumes or extract target chapters prior to uploading. ### 2. Physical File Size Exceeded In standard chat sessions, Claude standard chat conversations accept up to 20 files with a maximum file size of 500MB per file. Claude Projects accept files up to 30MB each with no fixed file count cap, bounded by context window capacity. If you attempt to drag a `35 MB` research archive into a project, the interface rejects the file immediately. For projects, compressing files or converting them from PDF to plain text (.txt or .md) typically brings them well below the Claude Projects 30MB file size limit. ### 3. Context Window Saturation Claude occasionally rejects or aborts processing on files that pass the physical size check. A dense dataset containing hundreds of thousands of rows of numerical data fits within the Claude standard chat 500MB file limit, but converting that tabular data into text tokens can produce `500,000` tokens. When a prompt with attached files exceeds the `200,000`-token context capacity, the conversation returns an error indicating that the message is too long to process. ### 4. Image Dimension Limits Images uploaded to Claude must not exceed `8000 x 8000` pixels. Anthropic recommends using images of at least `1000 x 1000` pixels for optimal OCR and chart interpretation, but ultra-high-resolution panoramic scans or satellite imagery exceeding `8,000` pixels on either axis will trigger upload rejections. ### 5. Disabled Code Execution for Spreadsheets Attempting to upload an Excel spreadsheet (.xlsx) without enabling analysis tools can cause ingestion failures. Claude requires code execution and file creation to be active in your account settings to parse and execute Python operations on `.xlsx` workbooks. If this setting is disabled, convert the spreadsheet to CSV or plain text before attaching it. ## What File Formats and Media Types Does Claude Support? Claude provides native support for `10` document formats and `4` image formats. Understanding these specifications prevents failed uploads and data loss. ### Supported Document Formats Claude accepts the following document types across web chats and project files: - **PDF (.pdf):** Up to `1,000` pages. Analyzes visuals up to `100` pages; text-only from `101` to `1,000` pages. - **Microsoft Word (.docx):** Text extracted directly; embedded diagrams and images are not read. - **Plain Text (.txt):** Ingested cleanly with minimal token overhead. - **Comma-Separated Values (.csv):** Structured tabular data. - **JavaScript Object Notation (.json):** Parsed as raw text or structured data payloads. - **HTML (.html):** Ingests page markup, removing scripts and external assets. - **OpenDocument Text (.odt):** Open standard word processing files. - **Rich Text Format (.rtf):** Formatted text documents. - **EPUB (.epub):** Digital publications and e-books. - **Excel (.xlsx):** Supported when code execution and file creation are enabled in account settings. ### Supported Image Formats For visual interpretation, diagram reading, and OCR, Claude accepts: - **JPEG (.jpeg, .jpg)** - **PNG (.png)** - **GIF (.gif):** Static image analysis only; animated frames are not parsed dynamically. - **WebP (.webp)** Images must not exceed `8000 x 8000` pixels. ### Unsupported File Types and Media Claude cannot accept audio files (.mp3, .wav), video files (.mp4, .mov), executable binaries (.exe, .dmg), or compressed archives (.zip, .tar.gz). If you attempt to upload a `.zip` archive containing project documentation, Claude will reject the upload. You must uncompress the archive and attach individual documents. ### Page Referencing Rule for PDFs When prompting Claude about specific content inside an attached PDF, always reference the page number displayed in your PDF viewer rather than the page number printed on the document footer. Claude maps tokens to the physical sequence of pages in the file container. If a document includes `12` introductory pages numbered with roman numerals, printed page 1 corresponds to viewer page 13. ## How to Query Large Document Collections in Claude Without Hitting Limits When your team manages large document libraries, design assets, or extensive code repositories, repeatedly splitting files or hitting context window caps slows down research. Attaching dozens of documents directly to Claude clutters the conversation and consumes thousands of tokens before you write a prompt. Teams typically evaluate two main paths for handling large collections: 1. **Custom Vector Databases and Retrieval Scripts:** Developers can build custom pipelines using tools like Chroma, Pinecone, or LangChain. This approach provides precise control over document chunking, but requires maintaining embedding models, database infrastructure, and custom API glue code. 2. **Standard Cloud Storage Folders:** Teams often attempt to link standard Google Drive, Dropbox, Box, or OneDrive folders directly to their assistants. While these services store large files, they function as general file storage rather than indexed AI workspaces, lacking automated hybrid search and structured extraction. ### External Workspaces with Remote MCP Search A more maintainable solution is using an external workspace platform built for AI teams. Fastio provides intelligent workspaces where large document collections are stored, indexed, and retrieved on demand by Claude through the Model Context Protocol (MCP). Instead of attaching files directly to Claude, your team uploads the documents into a Fastio workspace. Fastio supports single-file upload limits that far exceed native chatbot boundaries: - **Starter Plan:** Up to `25 GB` single-file upload limit, `250 GB` storage, 3 seats, 5 workspaces, and 100,000 monthly credits. - **Business Plan:** Up to `50 GB` single-file upload limit, `5 TB` storage, 10 seats, 50 workspaces, and 600,000 monthly credits. - **Enterprise Plan:** Up to `100 GB` single-file upload limit, `25 TB` storage, 30 seats included, 200 workspaces, and 3,000,000 monthly credits. Each plan is billed monthly: Starter at `$9.99`, Business at `$49.99`, and Enterprise at `$199.99`. Annual plan options are also available for organizations that prefer yearly billing. Monthly plans start with a 14-day free trial that requires a credit card. Credits meter AI work. Storage allowances and member seats come directly with each plan subscription. Cloud Sync connects Google Drive, Dropbox, Box and OneDrive folders to a workspace, one-way or two-way, on a schedule or on demand. ### Connecting Claude to Fastio via Remote MCP When you enable Intelligence Mode on a Fastio workspace, the platform automatically indexes incoming documents for hybrid search, combining full-text keyword matching, semantic meaning, and search-by-metadata-value. Claude connects to Fastio through its remote MCP server over Streamable HTTP. Connecting Claude to your workspace requires just a few steps: 1. In Claude (web, desktop, or mobile), open **Customize**, then select **Connectors**. 2. Add a custom connector and enter `https://mcp.fast.io/mcp/tools` as the server URL. 3. Select **Connect** and complete the browser sign-in to authenticate your Fastio account. 4. During authentication, select the organizations and workspaces Claude can access. 5. In your chat window, open the **+** menu under Connectors and enable Fastio. Once connected, Claude uses consolidated MCP tools to search the workspace index. When you ask a question about your documents, Claude queries the workspace, retrieves only the relevant paragraphs, and cites the source file. The 500MB chat limit and 30MB project limit remain untouched, but your assistant gains access to gigabytes of research without filling its `200,000`-token context window. For structured extraction from forms, invoices, and contracts, [Metadata Views](/product/document-data-extraction/) turn workspace documents into filterable spreadsheets, giving human teammates and AI assistants structured data on demand. ## Frequently asked questions ### What is the maximum file size you can upload to Claude? The maximum file size for a direct upload in a standard Claude chat conversation is 500MB per file. In Claude Projects, individual files are capped at 30MB each. Developers using the Claude Files API can upload files up to 500MB each. ### How many files can you upload to a Claude Project? There is no fixed limit on the number of files you can upload to a Claude Project. Claude Projects accept files up to 30MB each with no fixed file count cap, bounded by context window capacity. You can upload dozens or hundreds of files, provided each file stays within the 30MB limit and the cumulative text fits within Claude's context window. ### Why does Claude say file too large when uploading? Claude displays an 'Uploaded file is too large' error when a file exceeds the 500MB chat limit or 30MB project limit, when a PDF exceeds `1,000` pages, when an image exceeds `8000 x 8000` pixels, or when extracted document text exceeds Claude's `200,000`-token context capacity. ### How do you search large document collections in Claude without uploading them? You can connect Claude to an external intelligent workspace like Fastio using the Model Context Protocol connector at `https://mcp.fast.io/mcp/tools` or explore /storage-for-agents/. Files stored in the workspace are indexed automatically, allowing Claude to search and retrieve specific excerpts with citations on demand without exhausting chat context. ### Does Claude support Excel spreadsheets (.xlsx)? Yes, Claude supports Excel spreadsheets (.xlsx), provided code execution and file creation are enabled in your account settings. If code execution is disabled, convert your workbook into a CSV or plain text file before uploading. ### What is the maximum page count for PDF uploads in Claude? The maximum length for a PDF upload in Claude is `1,000` pages. Documents over `1,000` pages are rejected automatically. For PDFs of `100` pages or fewer, Claude analyzes both text and visual elements. For PDFs between `101` and `1,000` pages, Claude processes text only. ## Sources - [Anthropic Help Center: Upload files to Claude](https://support.claude.com/en/articles/8241126-upload-files-to-claude): Claude standard chat conversations accept up to 20 files with a maximum file size of 500MB per file. - [Anthropic Help Center: Upload files to Claude](https://support.claude.com/en/articles/8241126-upload-files-to-claude): Claude Projects accept files up to 30MB each with no fixed file count cap, bounded by context window capacity. ## 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.