MinerU Tianshu
Transforms various document formats like PDF, Office, and images into Markdown using a high-performance, multi-GPU service, and integrates with AI assistants via the Model Context Protocol (MCP).
About
MinerU Tianshu is an enterprise-grade document parsing service engineered for high performance and scalability. It features a modern, intuitive web interface built with Vue 3, TypeScript, and TailwindCSS, backed by a robust FastAPI backend. The service leverages LitServe for advanced GPU load balancing and efficient multi-GPU task handling, ensuring optimal processing for demanding document conversion tasks. With powerful parsing capabilities, it converts diverse file types, including PDFs, images, Word, Excel, PowerPoint, and more, into structured Markdown format. A standout feature is its native support for the Model Context Protocol (MCP), enabling seamless integration with AI assistants like Claude Desktop, allowing them to directly utilize its document parsing functionalities. Additionally, it offers comprehensive task management features, including task queues, prioritization, real-time status tracking, and automatic retries, making it ideal for enterprise environments.
Key Features
- High-performance architecture with GPU load balancing and multi-GPU isolation (LitServe)
- Comprehensive task management including queues, priorities, status tracking, and auto-retry
- Enterprise-grade multi-GPU document parsing (PDF, Office, image to Markdown)
- 25 GitHub stars
- Seamless AI assistant integration via Model Context Protocol (MCP)
- Modern web UI for task submission, monitoring, and management (Vue 3, FastAPI)
Use Cases
- Automating the conversion of large volumes of enterprise documents into structured Markdown.
- Integrating advanced document parsing capabilities directly into AI assistant workflows for enhanced content understanding.
- Building intelligent content management systems that require standardized Markdown from various document types.
