Document AI🔴Developer

Marker

Name: Marker
Brand: Marker
Price: 4 USD
Availability: InStock

High-performance open-source tool that converts PDFs, images, PPTX, DOCX, XLSX, HTML, EPUB, and other documents to markdown, JSON, chunks, or HTML with deep-learning-powered OCR, layout detection, and optional LLM cleanup.

Starting atFree

Visit Marker →

💡

In Plain English

Converts PDFs and documents to clean markdown or JSON — fast, accurate, handles tables, equations, and complex layouts with AI.

Overview

Marker is a free open-source document conversion pipeline for permitted research, personal, and qualifying startup use, with Datalab managed API pricing at $4 per 1,000 pages for Fast/Balanced mode, $6 per 1,000 pages for High Accuracy and extraction workflows, and custom self-hosting pricing. Marker is designed to turn complex documents into clean, structured outputs for AI, search, analytics, and knowledge-base workflows. Its core job is converting PDFs and other document formats into markdown, JSON, chunks, or HTML while preserving useful document structure such as headings, tables, forms, equations, inline math, links, references, images, and code blocks. Five concrete implementation facts define the tool: it is installable as the Python package marker-pdf; the README requires Python 3.10+; non-PDF support is enabled through the fuller marker-pdf[full] installation; the project supports local execution through CLI commands such as markersingle and marker; and it can also be used through Python APIs, a Streamlit GUI, or a lightweight FastAPI server. The input coverage is broad for a document AI converter: PDF, image, PPTX, DOCX, XLSX, HTML, and EPUB files are all described as supported inputs, while markdown, HTML, JSON, and chunks are documented outputs. The chunk output is especially relevant for retrieval workflows because it flattens top-level blocks for easier ingestion into downstream RAG or search systems. Marker also includes specialized modes for narrower tasks, including table-only conversion, OCR-only conversion, and beta structured extraction. Its optional --usellm mode can connect to services such as Gemini, Google Vertex, Ollama, Claude, OpenAI-compatible endpoints, and Azure OpenAI to improve hard cases like table merging across pages, inline math, table formatting, and form value extraction. Local deployment is practical for developers who can manage PyTorch and model dependencies, but resource planning matters: the README states Marker may use about 5GB of VRAM per worker at peak and about 3.5GB on average. Licensing also matters. The repository states that the code is GPL-3.0 and that model weights are free for research, personal use, and startups under $2M in funding or revenue, while broader commercial licensing or removing GPL requirements requires Datalab commercial licensing. For teams that do not want to operate the stack directly, Datalab offers a managed API, high-volume batch arrangements, and commercial self-hosted or on-premise options with custom terms.

🦞

Using with OpenClaw

▼

Use Marker as a Python library or CLI tool within OpenClaw skills for document processing. Shell out to marker_single or import from marker.converters for programmatic use.

Use Case Example:

Process PDFs uploaded to OpenClaw into clean markdown for memory storage, RAG retrieval, or document analysis tasks.

Learn about OpenClaw →

🎨

Vibe Coding Friendly?

▼

Difficulty:intermediate

CLI tool with Python library — requires Python environment setup and understanding of document processing concepts.

Learn about Vibe Coding →

Was this helpful?

Editorial Review

Marker is a leading open-source document conversion tool, combining deep learning layout detection with OCR to produce clean markdown, JSON, chunks, or HTML from complex documents. Its LLM-enhanced mode can improve difficult cases such as tables, forms, and math. Main limitations are GPL licensing restrictions, resource requirements for local throughput, and the need to use Datalab’s paid API or commercial self-hosting for many production commercial scenarios.

Key Features

Deep Learning Layout Detection+

Uses Surya models for detecting document regions: text blocks, headers, tables, figures, equations, code blocks, page headers, and footers. Handles multi-column layouts and complex page structures with reading order detection.

Use Case:

Converting a two-column research paper into single-column markdown with correct reading order and section hierarchy.

High-Quality OCR via Surya+

Integrated Surya OCR engine optimized for document text recognition. Supports 90+ languages and handles mixed-language documents with higher accuracy than Tesseract for most document types.

Use Case:

Processing scanned technical documents in multiple languages where Tesseract OCR produces too many errors.

Table Recognition & Structured Output+

Detects tables and converts them to properly formatted markdown tables or structured JSON with column alignment. Handles simple and moderately complex table structures, with LLM-enhanced mode for merging tables across pages.

Use Case:

Converting a technical specification PDF with comparison tables into structured data where table relationships are preserved.

LLM-Enhanced Conversion Mode+

Optional --use_llm flag pairs Marker with Gemini, Claude, OpenAI, or Ollama models to improve table formatting, handle inline math, extract form values, and merge tables split across pages. Benchmarks in the README report higher accuracy when LLM assistance is enabled for relevant cases.

Use Case:

Processing complex financial reports where tables span multiple pages and inline calculations need accurate LaTeX conversion.

Multi-Format Input & Output+

Accepts PDF, image, PPTX, DOCX, XLSX, HTML, and EPUB files. Outputs markdown, JSON (structured), chunks (pre-segmented for RAG), or HTML. Extensible with custom processors for specialized formatting logic.

Use Case:

Building an ingestion pipeline that converts a mix of PowerPoint presentations, Word documents, and PDFs into chunked JSON for a vector database.

Managed API & On-Premise Deployment+

Datalab offers a hosted managed API with published page-based pricing and custom self-hosted deployments for teams that do not want to operate the open-source stack directly. Published managed pricing lists $4 per 1,000 pages for Fast and Balanced mode and $6 per 1,000 pages for High Accuracy mode, structured extraction, track changes, and spreadsheets.

Use Case:

A compliance team that processes thousands of regulatory PDFs monthly using the managed API or a commercial self-hosted deployment to avoid maintaining a custom document conversion service.

Pricing Plans

Plan 1

Free for permitted uses

Plan 2

$4 per 1,000 pages for Fast and Balanced mode; $6 per 1,000 pages for High Accuracy mode, structured extraction, track changes, and spreadsheets; $25 monthly credit included on the managed plan

Plan 3

Custom pricing

Plan 4

Custom pricing

See Full Pricing →Free vs Paid →Is it worth it? →

Ready to get started with Marker?

View Pricing Options →

Getting Started with Marker

1Install Marker with pip: pip install marker-pdf (add [full] for non-PDF format support)
2Ensure Python 3.10+ and PyTorch are installed (GPU recommended for batch processing)
3Convert a single PDF: marker_single input.pdf --output_dir ./output
4For higher accuracy on complex documents, add --use_llm flag with Gemini or Ollama
5For production workloads, consider the managed API at datalab.to for hands-off processing

Ready to start? Try Marker →

Best Use Cases

🎯

Building RAG knowledge bases from document collections: Converting academic papers, technical docs, and books into clean markdown or chunked JSON for vector database ingestion where document structure preservation matters

⚡

Processing research papers with complex layouts: Handling multi-column academic papers with equations, tables, figures, and citations that break simpler extraction tools like PyPDF or pdfminer

🔧

Batch document conversion for search indexes: Processing large document libraries (hundreds to thousands of files) into searchable markdown for documentation sites, internal wikis, or full-text search systems

🚀

Multi-format document ingestion pipelines: Teams processing a mix of PDFs, PPTX, DOCX, and EPUB files that need a single tool handling all formats with consistent high-quality output

Integration Ecosystem

15 integrations

Marker works with these platforms and services:

🧠 LLM Providers

GoogleOpenAIAnthropic

📊 Vector Databases

custom-rag-pipelines

☁️ Cloud Platforms

AWS

💬 Communication

not documented

📇 CRM

not documented

🗄️ Databases

not documented

🔐 Auth & Identity

api-key

📈 Monitoring

usage-analytics

🌐 Browsers

not documented

💾 Storage

local-files

⚡ Code Execution

Docker

🔗 Other

GitHubollama

View full Integration Matrix →

Limitations & What It Can't Do

We believe in transparent reviews. Here's what Marker doesn't handle well:

⚠Very complex document layouts, especially nested tables and forms, may not convert perfectly.
⚠Forms may not render well in some cases, although the README says --use_llm and --force_ocr mostly solve the known issues.
⚠The simple built-in API server is intended for small-scale use and is not presented as a robust production service.
⚠OCR-heavy and high-throughput workloads require careful resource planning because worker count increases CPU/GPU usage and VRAM demand.
⚠Structured extraction is labeled beta and requires an LLM service to be configured.
⚠Some accuracy features depend on external or local LLM configuration, so results and costs can vary depending on the selected service.

Pros & Cons

✓ Pros

✓Supports multiple input types beyond PDF, including images, PPTX, DOCX, XLSX, HTML, and EPUB, which makes it useful for heterogeneous document collections.
✓Outputs markdown, HTML, tree-structured JSON, and flattened chunks, giving teams practical formats for human review, downstream parsing, and RAG indexing.
✓Optional LLM mode can improve hard cases such as cross-page tables, inline math, table formatting, and form value extraction, instead of relying only on OCR and layout models.
✓Developer-friendly architecture exposes converters, processors, renderers, providers, schemas, and block objects, so teams can customize the pipeline rather than treat it as a black box.
✓Includes table-only, OCR-only, and beta structured-extraction converters, which lets users run narrower pipelines when full-document conversion is unnecessary.
✓Benchmark data in the README reports strong speed and accuracy versus Llamaparse, Mathpix, and Docling, including favorable overall PDF conversion scores and improved table results with --use_llm.

✗ Cons

✗Local setup requires Python 3.10+, PyTorch, and model dependencies; non-PDF formats require the fuller marker-pdf[full] installation.
✗High-throughput local processing can be resource intensive: the README states Marker may use about 5GB VRAM per worker at peak and 3.5GB on average.
✗The built-in FastAPI server is described by the project as simple and intended only for small-scale use, so production API deployments may need the hosted Datalab API or custom infrastructure.
✗Known limitations remain for very complex layouts, especially nested tables and forms, and forms may not render well without extra OCR or LLM assistance.
✗Commercial use is not a simple permissive open-source story: the code is GPL-3.0 and broader commercial licensing or removing GPL requirements requires paid licensing.

Frequently Asked Questions

What file types can Marker convert?+

Marker supports PDF, image, PPTX, DOCX, XLSX, HTML, and EPUB files. The README notes that non-PDF document support requires installing additional dependencies with marker-pdf[full].

What output formats does Marker produce?+

Marker can output markdown, HTML, JSON, and chunks. Markdown includes image links, formatted tables, LaTeX equations, fenced code blocks, and footnote superscripts; JSON exposes a tree-like block structure; chunks flatten top-level blocks for easier RAG indexing.

Does Marker use LLMs?+

LLM use is optional. With --use_llm, Marker can improve accuracy for cases such as table merging across pages, inline math, table formatting, and extracting values from forms. The README lists Gemini, Google Vertex, Ollama, Claude, OpenAI-compatible endpoints, and Azure OpenAI as supported LLM services.

Can Marker run locally?+

Yes. Marker can run locally through CLI commands such as marker_single and marker, through Python APIs, through a Streamlit GUI, or through a lightweight FastAPI server. It can run on GPU, CPU, or Apple MPS, with Torch device detection and override options.

Is Marker free for commercial use?+

Not for all commercial situations. The repository states that the code is GPL-3.0 and the model weights use a modified AI Pubs Open Rail-M license that is free for research, personal use, and startups under $2M funding or revenue. Broader commercial licensing or removing GPL requirements requires Datalab’s commercial licensing.

🔒 Security & Compliance

—

SOC2

Unknown

—

GDPR

Unknown

—

HIPAA

Unknown

—

SSO

Unknown

✅

Self-Hosted

Yes

✅

On-Prem

Yes

—

RBAC

Unknown

—

Audit Log

Unknown

✅

API Key Auth

Yes

✅

Open Source

Yes

—

Encryption at Rest

Unknown

—

Encryption in Transit

Unknown

Data Retention: not documented for the hosted platform; local and self-hosted deployments keep data in the user's environment

Data Residency: EU/AU DATA RESIDENCY AVAILABLE ON CUSTOM TERMS

🦞

New to AI tools?

Read practical guides for choosing and using AI tools

Read Guides →

Get updates on Marker and 370+ other AI tools

Weekly insights on the latest AI tools, features, and trends delivered to your inbox.

What's New in 2026

•The GitHub repository page lists v1.10.2 as the latest visible release in the enrichment snapshot.

•The README currently describes Datalab’s managed platform as running Chandra, its latest open-source model, with higher accuracy than Marker and custom BAAs, while retention and compliance details should be verified directly with Datalab before procurement.

•The README reports active benchmark positioning against Llamaparse, Mathpix, and Docling, including Marker’s overall PDF conversion scores and H100 throughput projections.

•The current README includes support for multiple LLM services in hybrid mode, including Gemini, Google Vertex, Ollama, Claude, OpenAI-compatible endpoints, and Azure OpenAI.

Alternatives to Marker

Docling

MCP / Agent Infrastructure

IBM-originated open-source document processing software for parsing, understanding, serializing, and chunking complex documents for AI pipelines.

LlamaParse

Document AI

LlamaParse: Extract and analyze structured data from complex PDFs and documents using LLM-powered parsing.

Unstructured

Document Processing & OCR

Unstructured data platform for GenAI that connects to any source, processes 64+ file types, and outputs clean AI-ready inputs.

Apache Tika

Automation & Workflows

Enterprise-grade text extraction and document processing framework that detects and extracts content from 1,000+ file formats. Free, containerized, and battle-tested across 18 years of production deployment.

Amazon Textract

Automation & Workflows

AWS document intelligence service that extracts text, tables, forms, and handwriting from scanned documents using machine learning — with specialized APIs for invoices, IDs, and lending documents.

View All Alternatives & Detailed Comparison →

User Reviews

No reviews yet. Be the first to share your experience!

Quick Info

Try Marker Today

Get started with Marker and see if it's the right fit for your needs.

Get Started →

Need help choosing the right AI stack?

Take our 60-second quiz to get personalized tool recommendations

Find Your Perfect AI Stack →

Want a faster launch?

Explore 20 ready-to-deploy AI agent templates for sales, support, dev, research, and operations.

Browse Agent Templates →

More about Marker

Pricing Review Alternatives Free vs Paid Pros & Cons Worth It?Tutorial

📚 Related Articles

Best AI Tools for Document Processing & Data Extraction (2026)

A practical guide to AI-powered document processing tools. Compare Unstructured, LlamaParse, Amazon Textract, and more for extracting structured data from PDFs, invoices, contracts, and reports.

2026-03-1714 min read