UI-Venus-2 技術報告

摘要

多模態GUI代理已成為數位任務自動化的一個前景可期的典範，然而，由於環境覆蓋範圍有限、任務建構脆弱以及獎勵驗證不可靠，從以基準為導向的模型過渡到可靠的實際應用仍然充滿挑戰。在本研究中，我們提出了 UI-Venus-2，這是一個通用型基礎GUI代理，旨在透過統一的閉環推理-行動框架，在行動裝置、網頁和桌面環境中運行。為縮小通往實際部署的差距，我們聯合擴展了三個關鍵維度：(1) 環境——將覆蓋範圍擴展至逾170個多語言行動應用程式及原生桌面作業系統；(2) 任務——採用深度研究管線進行以功能為根基的指令生成；(3) 驗證——採用具有視覺關鍵點與多模型投票的軌跡層級與樣本層級評估器，以確保為訓練提供可靠的強化學習訊號。此外，我們整合了安全感知機制，以確保高影響力行動的受控執行。透過提供一個功能強大、高效且開源的基礎，UI-Venus-2 推動該領域朝向更可泛化、可驗證且具自我反思能力的代理邁進，以因應實際世界的應用需求。

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 signals for training. Furthermore, we integrate safety-aware mechanisms to ensure controlled execution of consequential actions. By offering a capable, efficient, and open-source foundation, UI-Venus-2 advances the field toward more generalizable, verifiable, and self-reflective agents for real-world applications.