1 paper
Zhengyang Tang, Yuxuan Liu, Xin Lai +21
A central bottleneck for phone-use agents is that controllable, reproducible environments covering real mobile behavior are hard to build at scale. Existing mobile-agent benchmarks…