UI-Voyager achieves an 81.0% success rate on AndroidWorld, exceeding human-level performance in mobile GUI automation.
March 26, 2026
Original Paper
UI-Voyager: A Self-Evolving GUI Agent Learning via Failed Experience
arXiv · 2603.24533
The Takeaway
The model uses Rejection Fine-Tuning and Group Relative Self-Distillation to learn from failed trajectories without manual annotation. This represents a leap in agentic capability where models can self-evolve to handle long-horizon digital tasks better than humans.
From the abstract
Autonomous mobile GUI agents have attracted increasing attention along with the advancement of Multimodal Large Language Models (MLLMs). However, existing methods still suffer from inefficient learning from failed trajectories and ambiguous credit assignment under sparse rewards for long-horizon GUI tasks. To that end, we propose UI-Voyager, a novel two-stage self-evolving mobile GUI agent. In the first stage, we employ Rejection Fine-Tuning (RFT), which enables the continuous co-evolution of da