Tips & Troubleshooting
Practical advice for getting the most out of Treasure Studio.
Working with PDFs
Studio can process PDF files attached to chat messages — extract text, summarize content, answer questions about documents. Here's what to know for reliable results.
Limitations
| Constraint | Limit |
|---|---|
| Max file size | 32 MB |
| Max pages | 100 pages (varies by model) |
| Encrypted PDFs | Not supported — remove password protection before uploading |
| Token cost | ~1,500–3,000 tokens per page (reduces available context window) |
Dense PDFs — small fonts, complex tables, heavy graphics — may fail even under page limits. The AI needs to convert each page into tokens, and visually complex pages consume more than simple text.
Best Practices
- Use text-based PDFs — Natively digital PDFs (not scanned images) give the best results. Scanned documents may work but accuracy drops significantly.
- Keep PDFs under 50 pages — Processing is most reliable with shorter documents. For longer PDFs, split them into sections and upload the relevant part.
- Remove password protection — Encrypted or password-protected PDFs cannot be processed. Remove protection before uploading.
- Mind the context window — A 50-page PDF can consume 75,000–150,000 tokens. Check the context window indicator to see how much space remains. Use a project to carry context across multiple chats if you're working with large documents.
- Use Opus for complex documents — Dense tables, charts, and multi-column layouts benefit from Opus's stronger reasoning capabilities.
Troubleshooting: "Could not process PDF"
If you see a "Could not process PDF" error, try these steps:
- Check the file size — Must be under 32 MB.
- Check the page count — Reduce to under 100 pages. For best results, keep under 50.
- Verify the PDF isn't encrypted — Open it in a PDF reader to confirm no password is required.
- Try a simpler version — If the PDF has heavy graphics or complex formatting, try exporting a text-only version from the source application.
- Convert to images — As a last resort, convert individual pages to PNG/JPEG and attach those instead. This bypasses PDF parsing entirely.
- Split the document — Extract only the pages you need using a PDF tool (Preview on macOS, or any PDF editor).
TIP
For the latest details on supported PDF features, see Anthropic's PDF support documentation.