Enterprise-GradePDF Parser For AI Workflow.
Cloud API & private deployment
Upload Documentcurl https://nanodoc.ai/nanodoc/api/v1/parse \
--request POST \
--header 'Authorization: Bearer {API_KEY}' \
--form 'file=@/path/to/file.pdf;type=application/pdf'Messy PDFs, Accurate Output.
We benchmark leading document parsing tools including NanoDoc, PaddleOCR, MinerU, GLM-OCR and LiteParse. With internally annotated documents of various types, we horizontally compare accuracy on text, tables and formulas, and evaluate overall parsing performance.
Text
Table
Formula
One parser. Three ways to run it.
Start with our managed API in minutes. Or let your AI agent drive the same API through a skill. Move to your own infrastructure when data residency, volume or network policy requires it — the parsing output stays identical.
Free during public beta
Get an API key and integrate the same day, with no infrastructure to provision.
- 1,000 high-priority pages every account, every day
- Beyond that, requests still run — queued at low priority, never rejected
- Allowance resets daily; automatic model and quality updates
Let your agent do the parsing
Add NanoDoc to your AI agent as a skill. Describe what you need in plain language — your agent calls the NanoDoc API and hands back structured output, ready to use.
- Works with any agent that supports skills
- Plain-language requests, structured output back
- Same engine, same quality as the API
Your infrastructure, your rules
Deploy NanoDoc in your own VPC or on-premise environment under a commercial license.
- Documents never leave your network boundary
- Air-gapped and offline installations supported
- Volume licensing with dedicated support