The field of multimodal GUI understanding is moving towards more effective and efficient methods for grounding and reasoning in graphical user interfaces. Recent developments have focused on improving the ability of models to understand and interact with complex GUI environments, including the use of iterative reasoning and reference feedback to improve grounding accuracy. Another key area of research is the development of more robust and generalizable GUI agents, which can navigate and interact with complex GUI environments through the use of multi-turn reinforcement learning and adaptive feature renormalization. Notable papers in this area include ChartPoint, which proposes a method for grounding MLLMs with reflective interaction and chain-of-thought reasoning, and Chain-of-Ground, which presents a training-free multi-step grounding framework that uses iterative visual reasoning and refinement to improve localization accuracy. Other notable papers include MPR-GUI, AFRAgent, and HiconAgent, which propose new methods for benchmarking and enhancing multilingual perception and reasoning in GUI agents, achieving high-resolution awareness in GUI automation, and optimizing history context-aware policy optimization for GUI agents, respectively.