Google just tested a bunch of new AI models for Android app coding – here are the rankings

Google refreshed Android Bench, its leaderboard for Android app coding models, and added latency, token usage, and cost metrics.

GPT 5.5 took first place, ahead of GPT 5.4 and Gemini 3.1 Pro, making it the top-ranked model for Android app development.

The update also brought several open-weight models onto the chart, including Claude Opus 4.7, GLM 5.1, Kimi K2.6, Gemma, Qwen, DeepSeek, and MiMo.

The benchmark underscores the tradeoff between coding quality and efficiency, since top performers may deliver better results but at higher latency and cost.