Google's Android bench ties GPT-5.4, Gemini 3.1 Pro preview 72.4%
Google just refreshed its Android Bench leaderboard, which ranks AI models used in Android app development.
OpenAI's brand-new GPT-5.4 and Google's own Gemini 3.1 Pro Preview now share the No. 1 spot, both scoring 72.4%.
The benchmark checks how well these AIs handle tasks like UI coding with Jetpack Compose and async programming using Coroutines.
OpenAI's GPT-5.3 Codex debuts 67.7%
OpenAI's GPT-5.3 Codex also made a strong debut, landing just behind the leaders with a score of 67.7%.
Other big names like Claude Opus 4.6 and older GPT models are still holding their ground in the rankings.
Google says these updates are all about helping developers build better apps faster, but reminds everyone to pick models that fit their specific project needs, since scores aren't everything.