The Year in AI Papers
The year/Independent research

Paper 2510.14528

PaddleOCR-VL: Boosting Multilingual Document Parsing via a 0.9B Ultra-Compact Vision-Language Model

Cheng Cui, Ting Sun, Suyin Liang, Tingquan Gao, Zelun Zhang, Jiaxuan Liu, Xueqing Wang, Changda Zhou, Hongen Liu, Manhui Lin, Yue Zhang, Yubo Zhang, et al.

Published
Oct 2025
Research lab
Independent
Citations
69
GitHub
Not linked

01 In brief

Summary

PaddleOCR-VL is a state-of-the-art, resource-efficient document parsing solution from Baidu's PaddlePaddle team.

It uses a two-stage architecture: PP-DocLayoutV2 for layout analysis (detection, classification, reading order) and PaddleOCR-VL-0.9B, a compact vision-language model combining a NaViT-style dynamic resolution visual encoder with the ERNIE-4.5-0.3B language model, for element recognition.

The model supports 109 languages and handles text, tables, formulas, and charts.

Training used over 30 million samples from public, synthetic, network, and in-house sources, with automatic annotation via large models and hard-case mining.

On OmniDocBench v1.5, PaddleOCR-VL achieves an overall score of 92.86, surpassing MinerU2.5 (90.67) and other baselines.

It also leads on OmniDocBench v1.0 (avg edit distance 0.115) and olmOCR-Bench (80.0% pass rate).

Element-level evaluations show top results in text, table (TEDS 0.9195 on OmniDocBench-Table-block), formula (CDM 0.9453 on OmniDocBench-Formula-block), and chart recognition (RMS-F1 0.8440 on in-house).

Inference is fast: with FastDeploy, it processes 1.6184 pages/s and 2486.4 tokens/s on an A100, outperforming competitors.

The model is open-sourced on GitHub and Hugging Face.

02 From the paper

Abstract

In this report, we propose PaddleOCR-VL, a SOTA and resource-efficient model tailored for document parsing. Its core component is PaddleOCR-VL-0.9B, a compact yet powerful vision-language model (VLM) that integrates a NaViT-style dynamic resolution visual encoder with the ERNIE-4.5-0.3B language model to enable accurate element recognition. This innovative model efficiently supports 109 languages and excels in recognizing complex elements (e.g., text, tables, formulas, and charts), while maintaining minimal resource consumption. Through comprehensive evaluations on widely used public benchmarks and in-house benchmarks, PaddleOCR-VL achieves SOTA performance in both page-level document parsing and element-level recognition. It significantly outperforms existing solutions, exhibits strong competitiveness against top-tier VLMs, and delivers fast inference speeds. These strengths make it highly suitable for practical deployment in real-world scenarios. Code is available at https://github.com/PaddlePaddle/PaddleOCR .