Mobile-R1

Vision language model-based mobile agents have gained the ability to not only understand complex instructions and mobile screenshots, but also optimize their action outputs via thinking and reasoning, benefiting from reinforcement learning, such as Group Relative Policy Optimization (GRPO). However, existing research centers on offline reinforcement learning training or online optimization using action-level rewards, which limits the agent's dynamic interaction with the environment. This often results in agents settling into local optima, thereby weakening their ability for exploration and error action correction. To address these challenges, we introduce an approach called Mobile-R1, which employs online reinforcement learning with task-level rewards for mobile agents. Our training framework consists of two stages: initial online training via action-level reward, followed by online training via task-level reward based on multi-turn trajectories. This strategy is designed to enhance the exploration and error correction capabilities of EurekaMove, leading to significant performance improvements. Moreover, we have collected a dataset with 24,521 high-quality manual annotations and established a new benchmark covering 28 Chinese applications and 1,510 instructions.