Lingshu

Welcome to the Lingshu project.

We show our SOTA multimodal large language models for the medical domain, Lingshu.
We release a unified evaluation framework, MedEvalKit, that consolidates major benchmarks for multimodal and textual medical tasks.

Highlights:

Lingshu supports more than 12 medical imaging modalities, including X-Ray, CT Scan, MRI, Microscopy, Ultrasound, Histopathology, Dermoscopy, Fundus, OCT, Digital Photography, Endoscopy, and PET.
Lingshu models achieve SOTA on most medical multimodal/textual QA and report generation tasks for 7B and 32B model sizes. Lingshu-32B outperforms GPT-4.1 and Claude Sonnet 4 in most multimodal QA and report generation tasks.
MedEvalKit allows for standardized, fair, and easy-to-use model assessment in medical domains, including multimdaol VQA, textual QA, and report generation.

Quick Links:

Models: Model weights are available in multiple model sizes: Lingshu-7B, Lingshu-32B
Evaluation Toolkit: We release our evaluation framework: MedEvalKit
Technique Report: Our technique report for the whole project is available at Lingshu Report .

Performance of Lingshu

Visual Question Answering for Various Medical Modalities

Medical Report Generation

Public Health

Medical Knowledge Understanding

Medical Visual Question Answering Results

Medical Textual Question Answering Results

Medical Report Generation Results

BibTeX

If you find our project useful, we hope you would kindly star our repo and cite our work as follows.
* `*` are equal contributions. `^` are corresponding authors.

@article{xu2025lingshu, title={Lingshu: A Generalist Foundation Model for Unified Multimodal Medical Understanding and Reasoning}, author={Xu, Weiwen and Chan, Hou Pong and Li, Long and Aljunied, Mahani and Yuan, Ruifeng and Wang, Jianyu and Xiao, Chenghao and Chen, Guizhen and Liu, Chaoqun and Li, Zhaodonghui and others}, journal={arXiv preprint arXiv:2506.07044}, year={2025} }