PulseAugur
EN
LIVE 03:45:25

Datalab's Marker 2 document converter achieves 5x speedup over MinerU

Datalab has released Marker 2, an open-source document conversion pipeline that supports various file formats including PDF, images, and Office documents, converting them into markdown, JSON, or HTML. The new version boasts significant improvements in speed and accuracy, achieving a 76.0% score on the olmOCR-bench benchmark and a throughput of 2.9 pages per second on a single GPU, which is over five times faster than its predecessor, MinerU. Marker 2 offers three conversion modes: balanced for high quality, fast for efficiency, and a disable_ocr option for pure text extraction, with enhanced CPU support and architectural changes to optimize performance. AI

IMPACT Improves document processing efficiency and accuracy for various file types, potentially speeding up data extraction workflows.

RANK_REASON This is a product release for an open-source tool, not a frontier model release or significant industry event.

Read on MarkTechPost →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Datalab's Marker 2 document converter achieves 5x speedup over MinerU

COVERAGE [1]

  1. MarkTechPost TIER_1 English(EN) · Asif Razzaq ·

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

    <p>Datalab rewrote Marker as a three-mode pipeline. Version 2 hits 76.0 on olmOCR-bench and sustains 2.9 pages per second on one B200 — over 5× MinerU's pipeline backend, while beating Docling on both accuracy and speed. Here's how it compares against MinerU, Docling and LitePars…