diff --git a/readme.md b/readme.md
index f1f1d6b..83b68e4 100644
--- a/readme.md
+++ b/readme.md
@@ -2,7 +2,7 @@
 
 <div align="center">
 
-[![arXiv](https://img.shields.io/badge/arXiv-2511.XXXXX-b31b1b.svg)](https://arxiv.org/abs/2502.XXXXX)
+[![arXiv](https://img.shields.io/badge/arXiv-2511.15831-b31b1b.svg)](https://arxiv.org/abs/2511.15831)
 [![Hugging Face](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Models-yellow)](https://huggingface.co/zwplus/UniFit)
 [![License](https://img.shields.io/badge/License-CC%20BY--NC--SA%204.0-lightgrey.svg)](./LICENSE)
 
@@ -75,3 +75,17 @@ Coming soon. We will provide weights for both SD3.5 and Flux.1 based models.
 Our code is modified based on [Diffusers](https://github.com/huggingface/diffusers). We use [Stable Diffusion 3.5 Medium](https://huggingface.co/stabilityai/stable-diffusion-3.5-medium) and [
 FLUX.1-Fill-dev](https://huggingface.co/black-forest-labs/FLUX.1-Fill-dev) as the base model. We adopt [
 Qwen2-VL-2B-Instruct](https://huggingface.co/Qwen/Qwen2-VL-2B-Instruct) as the MGSA Module. Thanks to all the contributors!
+
+
+## 📝 Citation
+
+If you find UniFit useful for your research, please cite our paper:
+
+```bibtex
+@article{zhang2025unifit,
+  title={UniFit: Towards Universal Virtual Try-on with MLLM-Guided Semantic Alignment},
+  author={Zhang, Wei and Jin, Yeying and Li, Xin and Zhang, Yan and Cong, Xiaofeng and Wang, Cong and Qiao, Fengcai and others},
+  journal={arXiv preprint arXiv:2511.15831},
+  year={2025}
+}
+```