Google Expands Android Bench with New LLMs, but Gemini Still Lags Behind

Google has updated its Android Bench benchmark for evaluating large language models on Android development tasks, adding eight new models such as Claude Fable 5, Claude Sonnet 5, and Qwen 3.7 Max. The benchmark now includes metrics for cost and efficiency, and a new framework designed to be easier for developers to use. Despite the additions, Google's own Gemini models continue to trail behind the new competitors on the leaderboard.