AI ToolsJul 25, 2026, 2:14 AM

Datalab’s Marker 2 vs MinerU, Docling and LiteParse: 76.0 on olmOCR-bench at 5× MinerU’s Throughput

30-second summary

Datalab has released Marker 2, an updated OCR pipeline achieving 76.0 on the olmOCR-bench and processing documents five times faster than MinerU on a B200 GPU, also outperforming Docling in both speed and accuracy.

TickrWire
Datalab’s Marker 2 vs MinerU, Docling and LiteParse: 76.0 on olmOCR-bench at 5× MinerU’s Throughput
Key takeaways
  • Marker 2 achieves 76.0 on the olmOCR-bench, indicating high accuracy in OCR tasks.
  • It boasts a throughput of 2.9 pages per second on an NVIDIA B200, significantly faster than competitors like MinerU.
  • The updated OCR pipeline is designed with three modes, offering improved efficiency and performance.
  • Marker 2 aims to provide a superior balance of speed and accuracy compared to existing solutions like Docling and LiteParse.
Full story

Datalab's Marker 2 represents a significant advancement in optical character recognition technology, featuring a re-engineered three-mode pipeline. This new version demonstrates a substantial performance leap, achieving a score of 76.0 on the industry-recognized olmOCR-bench, a key metric for OCR accuracy.

A major highlight of Marker 2 is its processing speed, capable of handling 2.9 pages per second on a single NVIDIA B200 GPU. This throughput is reported to be over five times faster than MinerU's pipeline backend, positioning Marker 2 as a highly efficient solution for large-scale document digitization.

Furthermore, Datalab claims Marker 2 surpasses Docling, another prominent OCR tool, in both accuracy and operational speed. The article provides a detailed comparison against MinerU, Docling, and LiteParse, helping users determine the best fit for their specific document processing needs.

Sponsored
Why this matters
Developers

Offers a potentially faster and more accurate OCR tool for integration into applications requiring document processing.

Businesses

Provides a high-throughput solution for digitizing large volumes of documents, potentially reducing operational costs and time.

Investors

Highlights innovation in the document AI space, indicating potential for market disruption and growth in enterprise solutions.

Everyone

Advances the capability of AI to accurately and quickly convert physical documents into editable digital text.

Glossary
OCR
Optical Character Recognition, technology that converts different types of documents, such as scanned paper documents, PDFs, or images captured by a digital camera, into editable and searchable data.
olmOCR-bench
A benchmark specifically designed to evaluate the performance and accuracy of Optical Character Recognition (OCR) systems.
Throughput
The rate at which a system or component can process data or complete tasks, often measured in units per second or per unit of time.
Sources · 1
Read next
More stories
TickrWireAI News Intelligence

We aggregate, verify, summarise and explain the latest artificial intelligence news from open, legal sources.

Daily AI digest

Top AI stories, summarised, in your inbox each morning.

© 2026 TickrWire. Summaries and analysis are AI-generated and may contain errors.