Google Rolls Out Prompt API for On‑Device Gemini and Agentic Android Studio Upgrades

The release focuses on faster Android development through on‑device privacy with scalable cloud image workflows.

Overview

Google introduced the ML Kit GenAI Prompt API in Alpha to let apps run custom prompts on Gemini Nano entirely on device.
Firebase AI Logic now exposes Gemini 2.5 Flash Image (“Nano Banana”) for image generation and leverages Imagen for mask-based edits, with multimodal processing across text, audio, and images.
Android Studio gains agentic features such as Agent Mode for multi-file changes, API upgrade assistance, a project assistant, and support for using any LLM to power IDE AI.
Google Play Console adds Gemini-powered app strings localization and AI-generated analytics summaries within a redesigned goal-oriented dashboard.
Google is finalizing an Android-specific LLM benchmark built from real GitHub repositories, with public results expected in the coming months.