UI-Venus Technical Report: Building High-performance UI Agents with RFT

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Gu, Zhangxuan, Zeng, Zhengwen, Xu, Zhenyu, Zhou, Xingran, Shen, Shuheng, Liu, Yunfei, Zhou, Beitong, Meng, Changhua, Xia, Tianyu, Chen, Weizhi, Wen, Yue, Dou, Jingya, Tang, Fei, Lin, Jinzhen, Liu, Yulin, Guo, Zhenlin, Gong, Yichen, Jia, Heng, Gao, Changlong, Guo, Yuan, Deng, Yong, Guo, Zhenyu, Chen, Liang, Wang, Weiqiang
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915447288365056
author Gu, Zhangxuan
Zeng, Zhengwen
Xu, Zhenyu
Zhou, Xingran
Shen, Shuheng
Liu, Yunfei
Zhou, Beitong
Meng, Changhua
Xia, Tianyu
Chen, Weizhi
Wen, Yue
Dou, Jingya
Tang, Fei
Lin, Jinzhen
Liu, Yulin
Guo, Zhenlin
Gong, Yichen
Jia, Heng
Gao, Changlong
Guo, Yuan
Deng, Yong
Guo, Zhenyu
Chen, Liang
Wang, Weiqiang
author_facet Gu, Zhangxuan
Zeng, Zhengwen
Xu, Zhenyu
Zhou, Xingran
Shen, Shuheng
Liu, Yunfei
Zhou, Beitong
Meng, Changhua
Xia, Tianyu
Chen, Weizhi
Wen, Yue
Dou, Jingya
Tang, Fei
Lin, Jinzhen
Liu, Yulin
Guo, Zhenlin
Gong, Yichen
Jia, Heng
Gao, Changlong
Guo, Yuan
Deng, Yong
Guo, Zhenyu
Chen, Liang
Wang, Weiqiang
contents We present UI-Venus, a native UI agent that takes only screenshots as input based on a multimodal large language model. UI-Venus achieves SOTA performance on both UI grounding and navigation tasks using only several hundred thousand high-quality training samples through reinforcement finetune (RFT) based on Qwen2.5-VL. Specifically, the 7B and 72B variants of UI-Venus obtain 94.1% / 50.8% and 95.3% / 61.9% on the standard grounding benchmarks, i.e., Screenspot-V2 / Pro, surpassing the previous SOTA baselines including open-source GTA1 and closed-source UI-TARS-1.5. To show UI-Venus's summary and planing ability, we also evaluate it on the AndroidWorld, an online UI navigation arena, on which our 7B and 72B variants achieve 49.1% and 65.9% success rate, also beating existing models. To achieve this, we introduce carefully designed reward functions for both UI grounding and navigation tasks and corresponding efficient data cleaning strategies. To further boost navigation performance, we propose Self-Evolving Trajectory History Alignment & Sparse Action Enhancement that refine historical reasoning traces and balances the distribution of sparse but critical actions, leading to more coherent planning and better generalization in complex UI tasks. Our contributions include the publish of SOTA open-source UI agents, comprehensive data cleaning protocols and a novel self-evolving framework for improving navigation performance, which encourage further research and development in the community. Code is available at https://github.com/inclusionAI/UI-Venus.
format Preprint
id arxiv_https___arxiv_org_abs_2508_10833
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle UI-Venus Technical Report: Building High-performance UI Agents with RFT
Gu, Zhangxuan
Zeng, Zhengwen
Xu, Zhenyu
Zhou, Xingran
Shen, Shuheng
Liu, Yunfei
Zhou, Beitong
Meng, Changhua
Xia, Tianyu
Chen, Weizhi
Wen, Yue
Dou, Jingya
Tang, Fei
Lin, Jinzhen
Liu, Yulin
Guo, Zhenlin
Gong, Yichen
Jia, Heng
Gao, Changlong
Guo, Yuan
Deng, Yong
Guo, Zhenyu
Chen, Liang
Wang, Weiqiang
Computer Vision and Pattern Recognition
We present UI-Venus, a native UI agent that takes only screenshots as input based on a multimodal large language model. UI-Venus achieves SOTA performance on both UI grounding and navigation tasks using only several hundred thousand high-quality training samples through reinforcement finetune (RFT) based on Qwen2.5-VL. Specifically, the 7B and 72B variants of UI-Venus obtain 94.1% / 50.8% and 95.3% / 61.9% on the standard grounding benchmarks, i.e., Screenspot-V2 / Pro, surpassing the previous SOTA baselines including open-source GTA1 and closed-source UI-TARS-1.5. To show UI-Venus's summary and planing ability, we also evaluate it on the AndroidWorld, an online UI navigation arena, on which our 7B and 72B variants achieve 49.1% and 65.9% success rate, also beating existing models. To achieve this, we introduce carefully designed reward functions for both UI grounding and navigation tasks and corresponding efficient data cleaning strategies. To further boost navigation performance, we propose Self-Evolving Trajectory History Alignment & Sparse Action Enhancement that refine historical reasoning traces and balances the distribution of sparse but critical actions, leading to more coherent planning and better generalization in complex UI tasks. Our contributions include the publish of SOTA open-source UI agents, comprehensive data cleaning protocols and a novel self-evolving framework for improving navigation performance, which encourage further research and development in the community. Code is available at https://github.com/inclusionAI/UI-Venus.
title UI-Venus Technical Report: Building High-performance UI Agents with RFT
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.10833