StackMap
Subscribe

MinerU vs PixelRAG

Heavyweight document-to-markdown/JSON parser — PDFs plus Office (docx/pptx/xlsx) through layout analysis and OCR into LLM-ready output for RAG and agentic pipelines. 73k stars, self-hostable. — versus — Berkeley's visual RAG: render pages and PDFs to screenshot tiles and retrieve with a VLM embedder — tables, charts and layout survive. pixelshot CLI plus a hosted 8.28M-page Wikipedia index.

The curated verdict

Two answers to the same RAG-ingestion problem: MinerU parses PDFs/Office through layout analysis and OCR into LLM-ready markdown; PixelRAG skips parsing entirely and retrieves over rendered screenshot tiles. Parse-to-text vs stay-in-pixels.

MinerUPixelRAG
Stars75k6.9k
Forks6.3k580
LanguagePythonPython
LicenseNOASSERTIONApache-2.0
Last activity3 days ago4 days ago
Topicsocr, ragrag, vision
Curated connections72

MinerU — the curator's take

The incumbent when document variety is the problem: beyond PDFs it handles Office formats, with mature layout analysis (reading order, tables, formulas) and a huge user base shaking out edge cases. NOT the lightest option — it's a full pipeline with model downloads and real hardware appetite; for a handful of clean PDFs a smaller tool is faster to stand up. Check the license (NOASSERTION on GitHub — AGPL-family, matters for commercial use).

PixelRAG — the curator's take

The paper's claim — screenshots beat parsed text for RAG — matters when your answers live in visual structure: tables, charts, infographics, layout-heavy PDFs that text chunkers flatten into noise. Zero-setup on-ramp is real: a hosted 8.28M-page Wikipedia index with no API key, and the pixelbrowse Claude Code plugin gives the agent screenshot-based page reading (pixelshot via Playwright/CDP, no MCP server). NOT a drop-in for text RAG stacks: your reader must be a VLM, tile indexes cost more storage/compute than text embeddings, and retrieval quality rides on their LoRA-tuned Qwen3-VL-Embedding model. Research codebase — expect pipeline assembly, not a product.