Gemini 3.5 Flash: Unexpected Weakness in Android Coding Tests

Google's Android coding tests reveal an unexpected weakness in Gemini 3.5 Flash. The results show this new model trailing older versions despite its premium positioning. What's interesting is that Gemini 3.5 Flash missed the top five in the Android coding leaderboard. OpenAI's GPT 5.5 claimed first place with a score of 74, and even Google's own Gemini 3.1 Pro Preview outperformed its successor, both scoring 72.4. Claude Opus models also delivered better results. Gemini 3.5 Flash scored 63.7, placing sixth overall. It also proved to be the most expensive option, averaging $147.1 per run, despite its slower performance. This is puzzling because Google's Flash branding typically suggests speed and lower prices. The bottom line is that while Google claimed Gemini 3.5 Flash had robust coding capabilities, its performance on actual Android development tasks appears less than stellar, raising questions about whether newer is always better.