Android Studio Quail 4 Goes Stable: Android Skills and Gemma 4 Power the AI Coding Era

Android Studio Quail 4 has officially reached the stable channel, marking the final stable release of the entire Quail development cycle. Announced at the start of September, the release is built around one clear idea: making artificial intelligence genuinely useful inside the IDE rather than an afterthought bolted onto the toolchain. For Flutter developers who do most of their Android work inside Studio, the update delivers both a mature build pipeline and two headline AI capabilities worth knowing about.

The announcement came from Amman Fasil Asfaw, Product Manager for Android Studio, who positioned Quail 4 as the release that lets teams build premium apps with AI both efficiently and effectively. While the cycle has been visible across several preview and candidate builds since earlier in the year, this stable arrival is what most production teams will actually adopt. It turns months of preview-channel experimentation into a dependable baseline you can pin your projects to.

Android Skills Bring Instant Domain Expertise

The most distinctive new feature in Quail 4 is the arrival of Android skills bundled directly into the IDE. Android Studio now ships with 23 curated skills preloaded, giving the AI agent instant domain expertise and applying Google's own best practices without requiring long, manual setup prompts. Instead of teaching the assistant your preferences from scratch, you start with a foundation that already knows how Android and Compose projects are meant to be structured.

The practical benefit is that the agent can answer questions, suggest refactors, and generate code with a level of context that used to require feeding it extensive project documentation every time you started a session. Teams can now treat the bundled skills as a shared baseline and layer their own project-specific knowledge on top.

• 23 curated Android skills are preloaded and ready to use out of the box
• The agent applies Google's best practices with less manual setup overhead
• Skills cover Android, Compose, and related platform conventions

Gemma 4: Private and Offline AI Coding

The second headline feature is native integration of Gemma 4, which Google describes as its most powerful open model, for AI code assistance that runs locally on your machine. This is a direct response to the growing demand for private, secure, and offline coding assistance. Developers who prefer to keep their source code local no longer need to juggle a separate third-party local model setup.

System requirements are modest at the entry point. The smallest Gemma 4 models can run on machines with 12 GB of RAM, while systems with 32 GB or more deliver the best experience. That threshold places the feature within reach of many developer laptops and comfortable for most desktop workstations, which lowers the barrier for teams that have hesitated to rely on cloud-based assistance because of data privacy concerns.

• One-click management: pick Gemma in the Agent model selector and choose a model to download
• Manage models from Settings, then Tools, then AI, then Model Providers, then Gemma
• Your source code never leaves the local machine
• No token quota limits to worry about

For Flutter developers, the local model story is especially attractive because it means code assistance works even when you are offline, in a secure environment, or on a network where sending code to a cloud service is not acceptable. The guarantee that source never leaves your machine also removes one of the most common reasons teams are reluctant to let an AI agent look at their codebase in the first place.

A Smoother Agent Experience Overall

Quail 4 also refines the way you interact with the AI agent during multitasking. Building on the parallel chat support introduced in Quail 2, the stable release adds several UI enhancements designed to make AI interactions smoother, faster, and more transparent.

Hyperlinked code symbols are one of the most convenient additions. When the agent mentions a class name, function, method, or file path in its responses, that reference is now clickable, letting you jump straight to the definition rather than hunting through the codebase manually. It turns AI output into an interactive navigation surface instead of plain text.

The Recent Chats panel has also been upgraded so you can monitor the progress of parallel conversations in real time. When several agents are working on different tasks at once, you no longer need to guess which one has finished; the panel surfaces their status as they run, making it easier to keep multiple workflows moving at the same time.

Earlier in the Quail cycle, Google set the stage for today's baseline. App Quality Insights integration was one of the first features to land, pairing the IDE's quality data with the Gemini agent, and a built-in MCP Server manager lets you install and manage Model Context Protocol servers directly from the IDE.

For Flutter teams, the implications ripple across both the IDE experience and the underlying Android build pipeline. The mature stable baseline that Quail 4 represents means the Android Gradle Plugin versions Flutter tooling pins to will be thoroughly tested, while the agentic workflows for Firebase configuration and Android builds become more dependable for everyday work. The release caps a deliberate, patient development cycle, and that patience shows in the polish.