Google says these AI models are best for coding Android apps
9to5Google • • general
Key Points:
- Google has launched "Android Bench," a new leaderboard that ranks AI models based on their performance in Android app development tasks, including UI design with Jetpack Compose, asynchronous programming, persistence, and dependency injection.
- The benchmark evaluates AI tools on handling complex Android-specific challenges such as navigation migrations, Gradle configurations, SDK updates, and integration with features like camera, media, and foldable devices.
- Google's testing revealed Gemini 3.1 Pro Preview as the top-performing AI model with a 72.4% score, followed by Claude Opus 4.6 and OpenAI’s GPT 5.2 Codex, while Gemini 2.5 Flash scored the lowest at 16.1%.
- By publishing these rankings, Google
