How to Upload a PDF to Claude for Analysis (Step-by-Step)

To upload a PDF to Claude for analysis, open claude.ai, start a new conversation, click the paperclip (or "Add files or photos") icon in the message input bar, select your PDF from your device, and send a prompt describing what you want Claude to do. Claude reads both the text and visual elements—charts, diagrams, and embedded graphics—inside the file, making it useful for everything from quick document summaries to complex multi-file analysis.

What file types can Claude analyze?

Claude supports a wide range of formats beyond PDF. According to Anthropic's support documentation, supported formats include PDFs, images (PNG, JPG, GIF, WebP), plain text, Markdown, CSV, TSV, DOCX, XLSX, and code files. For most of these, the core upload-and-analyze workflow works across all plans—Free, Pro, Max, Team, and Enterprise.

One important nuance: XLSX upload and file creation (generating output files) require code execution to be enabled, which is available on paid plans. If you're on the Free plan and need spreadsheet analysis, uploading a CSV instead is a reliable workaround.

How do you upload a PDF to Claude on claude.ai?

1. Open claude.ai and start a new conversation or open an existing Project.
2. Click the paperclip or "Add files or photos" icon in the message input bar.
3. Select your PDF from your device—or drag and drop it directly into the chat window.
4. Wait for the file chip to appear in the input bar, which confirms the upload is complete.
5. Type your prompt describing what you want Claude to do with the file, then send.

You can also select multiple files at once (hold Shift or Cmd when selecting) to upload several PDFs in a single message—useful for side-by-side document comparison.

What's the difference between uploading to a chat versus a Project?

This is one of the most practical distinctions to understand before you start:

• Individual chat uploads are temporary. Files attached directly to a chat are only available for that conversation. Once the conversation ends, the file is gone.
• Project Files uploads are persistent. If you open a Project, navigate to its Files section, and upload a PDF there, it remains available across all conversations within that project.

If you're doing a one-off analysis—summarizing a single report, for example—attaching to a chat is fine. If you have a policy document, style guide, or dataset your team returns to repeatedly, upload it to a Project so you never have to re-upload it.

How do you get the most out of PDF analysis?

The prompt you write matters as much as the file you upload. Here are three patterns that consistently produce useful output:

1. Summarization

Upload a long report and ask Claude to extract the key findings in bullet points. For example:

"Summarize the key financial metrics and main takeaways from this report in bullet points."

This is the most common entry point for file uploads—turning a dense document into an actionable summary in under a minute.

2. Multi-document comparison

Upload two or more PDFs at once and ask Claude to compare them across specific dimensions. A procurement analyst, for instance, might upload three vendor proposals and ask:

"Compare these three vendor proposals across base price, setup fee, delivery timeline, SLA uptime guarantee, and support hours. Format as a Markdown table."

Multi-file comparison saves hours of manual cross-referencing and reduces the chance of missing critical differences buried in dense contract language.

3. Visual element extraction

Because Claude can analyze both text and visual elements such as charts, diagrams, and embedded graphics within uploaded files, you can ask it to describe or extract data from figures that a plain text extractor would miss entirely. This is particularly valuable for research papers, financial reports, and technical documentation.

What are the most common pitfalls when uploading PDFs?

Very large or dense PDFs

PDFs over 100 pages or with heavy embedded graphics can overflow the context window even before reaching the page limit. The fix: split large PDFs into sections (for example, by chapter), and reduce embedded image resolution before uploading if the file is graphics-heavy.

DOCX files with embedded images

When you upload a DOCX file, embedded images are not parsed visually. If your Word document contains charts or diagrams you need Claude to analyze, convert it to PDF first using Microsoft Word's export function or a tool like LibreOffice, then upload the PDF.

Low-resolution images inside PDFs

For text-heavy images—screenshots of code, forms, or scanned documents—use lossless formats like PNG where possible, and avoid heavy compression on images with fine text or small details. Claude's accuracy on visual content degrades with low-resolution source material.

Files disappearing after a conversation

As noted above, files attached to individual chats are temporary. If you find yourself re-uploading the same PDF repeatedly, move it to a Project's Files section instead.

When should you use the Files API instead of the claude.ai interface?

If you're a developer building automated pipelines—processing invoices, analyzing batches of contracts, or running repeated analysis on the same documents—the Files API is worth understanding. It lets you upload files server-side, store them, and reference them by a unique file ID across multiple API requests without re-uploading each time. Files up to 500 MB per file are supported.

For one-off API calls where you don't want to manage server-side storage, inline encoding in the API payload is simpler. But for any workflow that processes the same file more than once, the Files API approach is more efficient.

Is uploading PDFs to Claude available on the Free plan?

Yes. File uploads are available across all plans—Free, Pro, Max, Team, and Enterprise. The core upload-and-analyze workflow works for most file types on all plans. The main limitation on the Free plan is that features requiring code execution (such as XLSX upload and generating output files) are not available. For PDF analysis specifically, the Free plan works without restriction.

How does uploading a PDF compare to pasting text directly?

The short rule: if your document has any visual content, or if it's long enough that pasting would be awkward, upload the PDF. If it's a short block of plain text, pasting is faster. See Anthropic's PDF support documentation for additional technical details on how PDF content is processed.

Quick reference: uploading a PDF to Claude

1. Go to claude.ai and open a new chat (or a Project for persistent access).
2. Click the paperclip icon in the message input bar, or drag and drop your PDF into the window.
3. Confirm the file chip appears—this means the upload succeeded.
4. Write a clear prompt: specify what you want (summary, comparison, data extraction, proofreading).
5. Send and review. If the output is too broad, follow up with a more specific question.

For multi-file analysis, select all files in the same upload step so Claude can reason across them simultaneously rather than in separate conversations.