#88th of 13 in Data and scraping for AI
marker
Converts PDFs, Office files and images to markdown, JSON, HTML or chunks
- Stars
- 40.4k
- License
- Apache-2.0
- Last commit
- Oct 2026
- Last release
- Jul 2026
- Language
- Python
32,688 on 25 Mar 202640,445 on 14 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Marker is a Python library and CLI that converts PDF, image, PPTX, DOCX, XLSX, HTML and EPUB files to markdown, JSON, HTML or chunks, handling tables, equations, code blocks and image extraction. Layout, OCR and table recognition run through a surya VLM served by a local vLLM (Docker, NVIDIA) or llama.cpp server that starts automatically. An optional --use_llm mode adds Gemini, Claude, OpenAI-compatible, Azure, Vertex, OpenRouter or Ollama for higher accuracy.
Who it is for: Engineers converting PDFs and documents to markdown or JSON for LLM pipelines
Strengths
- Outputs markdown, JSON block tree with polygons, HTML or chunks, plus extracted images
- Fast mode uses text layer and rf-detr layout on CPU; --disable_ocr starts no server
- Batch CLI with worker pool, --skip_existing and chunk sharding across multiple machines
- Reports 76.0% on olmocr-bench in balanced mode, ahead of MinerU and docling
Weaknesses
- Model weights use a modified OpenRAIL-M license; commercial use above startup limits needs a paid license
- Balanced mode needs Docker with NVIDIA Container Toolkit, or llama.cpp on CPU and Apple Silicon
- Non-PDF formats require the marker-pdf[full] extra install
- README says the hosted Chandra model is more accurate than Marker
What it needs
- GPU optional
- Needs PyTorch, Python 3.10+, vLLM via Docker with NVIDIA Container Toolkit (GPU), llama.cpp llama-server (CPU / Apple Silicon), optional LLM service for --use_llm
- Models: surya VLM, Gemini, Claude, OpenAI-compatible, Azure
Also in Data and scraping for AI
See all 13| Rank | Project | Score |
|---|---|---|
| 1 | mineruDocument parser that converts PDFs, images and Office files to Markdown and JSON | 78 out of 100 |
| 2 | Crawl4AIPython web crawler that converts pages to LLM-ready Markdown and JSON | 77 out of 100 |
| 3 | doclingPython library and CLI that converts PDFs, Office files and more into structured text | 76 out of 100 |
| #4 | markitdownPython tool that converts PDF, Office, HTML and audio files to Markdown for LLMs | 71 out of 100 |
| #5 | FirecrawlWeb scraping and crawling API that returns LLM-ready markdown | 67 out of 100 |