Google LLC today rolled out an upgrade to the coding assistant in Android Studio, its toolkit for building Android apps.

Android Studio is a so-called IDE, or integrated development environment. It’s a kind of word processor in which developers can write the code for new apps. Android Studio also offers other features, including an emulator that makes it possible to run virtual Android handsets on a Windows machine and use them to test new code.

Last May, Google added an artificial intelligence coding assistant dubbed Studio Bot to the IDE. It was based on a large language model called PaLM 2 that the search giant announced the same month. Today’s update replaces Studio Bot with Gemini in Android Studio, an improved coding assistant based on a newer AI model.

Six months after the debut of PaLM 2 last year, Google introduced a new flagship LLM lineup. The Gemini series, as it’s called, includes three models with varying feature sets and hardware requirements. Android Studio’s upgraded chatbot is based on Gemini Pro, the midrange model in the series, which is designed to balance response quality with hardware-efficiency. 

“If you were already using Studio Bot in the canary channel, you’ll continue experiencing the same helpful and powerful features, but you’ll notice improved quality in responses compared to earlier versions,” Android Studio product manager Sandhya Mohan detailed in a blog post today.

Gemini in Android Studio can answer general questions about Android, such as how to integrate an app with the operating system’s location tracking feature. The chatbot is also capable of answering project-specific questions. A developer could, for example, request advice on how to remove duplicate records from a dataset an app uses to power some of its features.

Gemini in Android Studio can take on more complex tasks as well. According to Google, the chatbot is capable of rewriting code files in a new programming language and generating entire app interface sections. Gemini in Android Studio can also write documentation that explains how newly created code works. 

“Gemini is only available after you log in and enable it,” Mohan detailed. “You don’t need to send your code context to take advantage of most features. By default, Gemini in Android Studio’s chat responses are purely based on conversation history, and you control whether you want to share additional context for customized responses.”

The chatbot is based on Gemini 1.0 Pro, an LLM that Google introduced last December. Since then, the search giant has detailed a newer iteration of the model called Gemini 1.5 Pro. It’s possible the latter model will roll out to Android Studio in a future update given the fact that it features enhanced coding capabilities.

According to Google, Gemini 1.5 Pro outperformed the 1.0 version that Android Studio uses across 87% of the benchmarks its engineers evaluated. In one internal test, the search giant provided Gemini 1.5 Pro with a prompt containing more than 100,000 lines of code. It proved more adept both at generating explanations and suggesting improvements.

Image: Google