– Tops in benchmark test, including reading order, tables, and title inference.

– Offers a perfect local security environment with the hybrid engine that combines AI and direct extraction heuristic engine

– Four AI functions via free AI add-ons; OCR, Table AI, Chart and Formula AI

– Change to Apache-2.0 licensing from MPL-2.0

SEOUL, South Korea, March 13, 2026 /PRNewswire/ — Hancom, the South Korean software company behind the widely used Hangul word processor, has released OpenDataLoader PDF v2.0 — and the benchmark numbers back up the claim. In the company’s own internal testing, OpenDataLoader PDF outperformed competing open-source tools across reading order recognition, table extraction, and heading inference. Hancom has published the full benchmark dataset and reproducible code on its official GitHub repository, allowing developers to verify the results independently.

The headline engineering move is a hybrid extraction engine that pairs AI-based parsing with direct extraction. The practical upside: enterprises and developers get high-accuracy PDF data extraction that runs entirely on-premise, with no data leaving the local environment. For organizations handling sensitive documents — legal, financial, medical — that’s not a minor footnote.

Four Free AI Add-ons, Out of the Box

OpenDataLoader PDF PDF v2.0 includes the following four AI features as add-ons at no additional cost:

• OCR— improves text recognition on image-based and scanned PDFs
• Table Extraction— a lightweight AI model that handles merged cells and complex table structures with precision
• Formula Extraction— recognizes mathematical and scientific notation locally, without a cloud call
• Chart Analysis— converts chart visuals into natural-language descriptions

All four are built for compatibility with third-party open-source models, including Docling. Hancom is clear that no formal partnership or sponsorship is in place — the compatibility is purely technical, designed so developers can slot OpenDataLoader PDF into existing pipelines without rebuilding their stack.

Apache 2.0: Lowering the Barrier, Expanding the Ecosystem

The project has also shed its MPL 2.0 license in favor of Apache 2.0 — one of the most permissive open-source licenses available. The move directly reduces friction for commercial use, making it easier for global developers and enterprises to build on top of OpenDataLoader PDF without navigating license compatibility headaches. Hancom expects this to accelerate downstream business models including WebApp and SaaS applications built on the engine.

Ecosystem Expansion: LangChain Is In, More Integrations Coming

LangChain integration shipped in 2025. In 2026, Hancom is targeting Langflow, LlamaIndex, and Gemini CLI, plus MCP (Model Context Protocol) support for agentic AI workflows. The roadmap positions OpenDataLoader PDF as infrastructure for the autonomous AI agent era, not just a standalone parsing tool.

Later in 2026, a commercial AI add-on is planned — described as a concentration of Hancom’s proprietary document AI technology.

AI based auto-tagging to Tagged PDF: Start The Accessibility Play

Perhaps the most forward-looking item on the roadmap is PDF accessibility. With the European Accessibility Act (EAA) now in force, South Korea’s anti-discrimination legislation tightening, and accessibility regulations expanding globally, compliance has become a real operational burden for enterprises. Hancom says OpenDataLoader PDF will be the first open-source PDF tool to include AI-generated accessibility tagging — This will be the first open-source solution to provide a key step toward PDF/UA compliance.

What Hancom’s CTO Said

“OpenDataLoader PDF v2.0 has evolved into an open PDF data platform that anyone can freely use and build upon, through its AI hybrid engine and transition to Apache 2.0,” said Jihwan Jeong, CTO of Hancom. “With upcoming commercial AI add-ons and accessibility solutions, we aim to lead the global ecosystem — making PDF documents not only AI-ready, but accessible to everyone.”

OpenDataLoader PDF PDF v2.0 is available now. Source code, benchmark datasets, and documentation are published at the OpenDataLoader PDF official GitHub repository.