Google DeepMind launched Nano Banana 2 Lite and Gemini Omni Flash for developers. These lightweight models prioritize low latency and efficiency over raw parameter count. They target edge-device deployment where memory is constrained. Practitioners can now integrate these smaller footprints into mobile apps without sacrificing basic multimodal reasoning capabilities.