Paper Overview
Research Area: NLP Authors: Venus Team, Zhuohan Cai, Haoxing Chen Published: 2026-09-03 arXiv: 2509.00008
Abstract (English)
Multimodal GUI agents have emerged as a promising paradigm for digital task automation, yet transitioning from benchmark-oriented models to dependable real-world applications remains challenging due to limited environment coverage, brittle task construction, and unreliable reward verification. In this work, we present UI-Venus-2, a general-purpose foundation GUI agent designed to operate across mobile, web, and desktop environments through a unified closed-loop reasoning-action framework. To bridge the gap toward practical deployment, we jointly scale three critical dimensions: (1) Environments, expanding coverage to more than 170 multilingual mobile apps and native desktop operating systems; (2) Tasks, employing a deep-research pipeline for function-grounded instruction generation; and (3) Verification, adopting trace-level and sample-level evaluators with visual keypoints and multi-model voting to ensure reliable RL training signals. In addition, we integrate safety-aware mechanisms to ensure controlled execution of critical actions. By providing a capable, efficient, and open-source foundation, UI-Venus-2 advances the field toward more general, verifiable, and self-reflective agents for real-world applications.
Key Contributions
- Unified closed-loop reasoning-action framework spanning mobile, web, and desktop environments.
- Scaled environments: coverage of 170+ multilingual mobile apps plus native desktop operating systems.
- Function-grounded task generation via a deep-research pipeline.
- Reliable reward verification using trace-level and sample-level evaluators with visual keypoints and multi-model voting for RL training.
- Safety-aware mechanisms for controlled execution of critical actions.
- Open-source foundation aimed at general, verifiable, self-reflective GUI agents.