PaddleOCR-VL: Boosting Multilingual Document Parsing via a 0.9B Ultra-Compact Vision-Language Model
Summary
PaddleOCR-VL is a compact 0.9B vision-language model that achieves state-of-the-art performance in multilingual document parsing and element recognition by integrating NaViT-style dynamic resolution with the ERNIE language model.
View Cached Full Text
Cached at: 05/08/26, 08:42 AM
Paper page - PaddleOCR-VL: Boosting Multilingual Document Parsing via a 0.9B Ultra-Compact Vision-Language Model
Source: https://huggingface.co/papers/2510.14528
Abstract
PaddleOCR-VL, a vision-language model combining NaViT-style dynamic resolution and ERNIE, achieves state-of-the-art performance in document parsing and element recognition with high efficiency.
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 powerfulvision-language model(VLM) that integrates aNaViT-style dynamic resolution visual encoderwith theERNIE-4.5-0.3B language model to enable accurateelement 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 bothpage-level document parsingandelement-level recognition. It significantly outperforms existing solutions, exhibits strong competitiveness against top-tierVLMs, 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 .
View arXiv pageView PDFGitHub77.3kAdd to collection
Get this paper in your agent:
hf papers read 2510\.14528
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper17
#### PaddlePaddle/PaddleOCR-VL Image-Text-to-Text• 1.0B• Updated8 days ago • 10.1k • 1.6k
#### PaddlePaddle/PP-DocLayoutV2 Object Detection• UpdatedJan 29 • 10.5k • 29
#### unsloth/PaddleOCR-VL Image-Text-to-Text• 1.0B• UpdatedDec 9, 2025 • 147 • 17
#### lvyufeng/PaddleOCR-VL-0.9B Image-Text-to-Text• 1.0B• UpdatedOct 21, 2025 • 120 • 4
Browse 17 models citing this paper## Datasets citing this paper1
#### proxectonos/corpus_dominio_cientifico Preview• Updated16 days ago • 61
Spaces citing this paper17
Collections including this paper31
Similar Articles
PaddleOCR-VL-1.6: Expanding the Frontier of Document Parsing with Under-Optimized Region Refinement and Progressive Post-Training
PaddleOCR-VL-1.6 improves document parsing by identifying and refining under-optimized regions via targeted data optimization and progressive post-training, achieving state-of-the-art 96.33% on OmniDocBench v1.6.
PaddleOCR 3.5: Running OCR and Document Parsing Tasks with a Transformers Backend
PaddleOCR 3.5 adds a Transformers inference backend, enabling OCR and document parsing models like PP-OCRv5 and PaddleOCR-VL 1.5 to run seamlessly within the Hugging Face ecosystem.
MinerU2.5: A Decoupled Vision-Language Model for Efficient High-Resolution Document Parsing
MinerU2.5 is a 1.2B-parameter vision-language model that achieves state-of-the-art document parsing accuracy with high computational efficiency using a coarse-to-fine parsing strategy.
NaviDC-OCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
NaviDC-OCR is a unified vision-language framework that improves document parsing accuracy through deformation-aware learning and adaptive sampling, achieving state-of-the-art results on multiple benchmarks.
MonkeyOCRv2: A Visual-Text Foundation Model for Document AI
MonkeyOCRv2 is a visual-text foundation model for document AI, pretrained on a large corpus of 113 million images across 17 languages using joint image-to-text generation and pixel-level document reconstruction. It achieves state-of-the-art results on document parsing and understanding tasks, outperforming previous models with a smaller vision encoder.