Choose theme
AI Resources
MinerU
MinerU turns PDFs, scanned pages and Office documents into content that software and AI workflows can use. It combines document conversion with a local library for finding files, searching extracted content and returning to particular pages or blocks.
You can export a document for another application or let an agent read it progressively through the library. Python, command-line, WebUI and self-hosted API routes serve different ways of using the parser. Use this as a first read, not a recommendation. Open the original project before trusting details like terms, limits, privacy, cost, setup, or safety.
What it is
Parsing before indexing or answering
MinerU extracts and organizes document content, including text, tables, formulas and images. Its Python parser returns a ParseResult that can produce Markdown or a reusable JSON document representation. A separate application can use those outputs for indexing, retrieval or analysis.
Why it stands out
Choose the parsing work the file needs
Its Flash tier handles native document parsing and previews; Basic adds model-based OCR, tables and formulas. Standard combines small models with a vision-language model, while Advanced uses more inference computation. Those choices change the models and runtime a workflow needs.
Availability
Public source and several entry routes
The repository provides code, releases and setup guides. A local WebUI offers a visual starting point after installation; developers can use Python or an API. The project declares the MinerU Open Source License; see the main repository for current terms.
Why it matters
What makes it useful
A report can mix ordinary text with scanned pages, tables and formulas. Copying its visible paragraphs may leave out material a search application needs. MinerU provides a conversion stage for those document elements before a downstream system indexes or answers from them. RAGFlow documents MinerU as a selectable PDF parser, with RAGFlow connecting to a separately deployed MinerU service.
What to know
Where it fits
There are two command-line workflows. The mineru commands work through the document library and its cache; mineru-kit parse performs direct conversion without creating that library. For a whole-file Markdown export, the quickstart uses mineru-kit parse with an output file. The local Python parser also works without a document library and can save a result for another program to load.
Notable points
What stands out
A JSON response and a JSON document are different outputs here. With mineru parse, --json returns command status, content, locators and any continuation request. It is not the reusable MiddleJson document representation. The Python ParseResult export or the appropriate mineru-kit output supplies that representation. This matters when connecting the tool to an application: saving an agent-command response does not give the application the same input as saving the parsed document.
Before using
What to review
The current package supports Python 3.10 through 3.14. Optional inference engines have narrower platform and driver requirements. The base installation includes ONNX small-model processing and llama.cpp; the chosen tier still needs its relevant models, which may download on first use.
For PDFs, mineru parse starts with the first ten pages unless another range is selected. A response may also stop at its output budget and provide a continuation request. A short returned passage therefore need not cover the full file; mineru-kit parse defaults to all PDF pages.
Flash with native-text mode reads an existing PDF text layer without an OCR fallback for scanned pages. Scans need the appropriate OCR mode and models. A configured API on another machine can receive uploaded documents; local installation alone does not describe that service connection.
Try a representative file and compare the extracted text, table structure and selected page range with the original before building a larger workflow around the output.
Reader fit
Who may find it relevant
Relevant to developers preparing mixed document collections for search, and technically comfortable readers who want file conversion or an agent-readable document library. Someone trying a few files can begin with the WebUI; a team integrating parsing into an application will also need to choose outputs, models and deployment settings.
Editorial note
Why LifeHubber lists it
LifeHubber lists MinerU because its document library connects a search result to a specific page or block that an agent can read again. For someone asking follow-up questions about a report, that makes the extracted document something to navigate rather than one long paste. The same location can also be requested as an image, giving a concrete way to revisit the table or figure behind a passage.
Source links
Source materials
Reader note
Before relying on this entry
LifeHubber lists entries to help readers inspect AI projects, not to endorse them or prove they are safe, suitable, accurate, maintained, or right for a specific use. We do not verify every entry in depth. Before relying on anything listed, review the original materials, terms, privacy practices, limits, and risks that matter for your situation.
What to explore next
Put parsed documents into a knowledge workflow
After exploring MinerU conversion, see how RAGFlow combines document ingestion with retrieval and cited answers. Its documentation includes a MinerU PDF-parser connection through a separately deployed service.
More in Ecosystem
Keep browsing this category
Explore more AI ecosystem resources.
Laya
NandhaKishorM/laya
An early Apache-2.0 model family and Python runtime for bounded choice, score, and yes-or-no decisions, with English, multilingual, and task-specialized checkpoints plus a router that selects between them.
Docling
docling-project/docling
A document conversion toolkit with structured text, tables and layout, native chunking for retrieval, Markdown and JSON exports, and local model configuration.
Monty
pydantic/monty
A Rust interpreter for AI-written Python, with explicit host-function access, resumable execution state, and Python, JavaScript/TypeScript and Rust integration paths.
For project maintainers
Listed here? You can use the badge.
If you maintain a project with a current LifeHubber listing, you may add the optional “Listed on LifeHubber AI Resources” badge to its README, docs, or website. No introduction or permission request is needed.
