AndroidWorld
An Android interface automation benchmark and research framework introduced by Google DeepMind, used to evaluate the ability of AI models to execute tasks on real Android application interfaces.
Core Facts
Timeline (last 90 days)
mobile-use topped Google DeepMind's AndroidWorld benchmark, becoming the first project to achieve a 100% score on that benchmark.
Qwen3 27B's performance on computer operation benchmarks is: OSWorld Verified 84.3, WebArena Verified 64.8, and AndroidWorld 81.9.
All Facts (3)
Google DeepMind explored the possibility of controlling mobile devices through the Android Accessibility API.
75%Unverifiedmobile-use topped Google DeepMind's AndroidWorld benchmark, becoming the first project to achieve a 100% score on that benchmark.
50%UnverifiedQwen3 27B's performance on computer operation benchmarks is: OSWorld Verified 84.3, WebArena Verified 64.8, and AndroidWorld 81.9.
50%