GitHub has broadened the capabilities of its AI-driven coding assistant, Copilot, by adding support for multiple new AI models from leading providers like Anthropic, Google, and OpenAI. Additionally, the company introduced GitHub Spark, a groundbreaking tool for developing web applications through natural language input.
With this update, GitHub Copilot users can now choose from a variety of advanced generative AI models. Among the newly supported models are Google Gemini 1.5 Pro, OpenAI’s GPT-4o, OpenAI’s o1-preview and o1-mini, and Anthropic’s Claude 3.5 Sonnet. Initially, the expanded model selection is exclusive to GitHub Copilot Chat, but GitHub confirmed it would soon be accessible in other Copilot features.
This addition was announced on October 29, alongside the launch of GitHub Spark, a dedicated tool for creating web applications entirely through natural language prompts. GitHub aims to make app creation as simple as possible, opening doors for developers of all levels to bring their ideas to life without extensive coding experience.
For users working in environments like GitHub.com or Visual Studio Code, Copilot now allows model switching within Copilot Chat to suit the task at hand. Alternatively, developers can rely on Copilot’s default model for general use. GitHub emphasizes this multi-model approach to provide a selection of top-tier models tailored to each user’s needs. Anthropic’s Claude 3.5 Sonnet is immediately available, while Google Gemini 1.5 Pro will roll out shortly. Access to the latest OpenAI models has also been activated.
GitHub Spark stands out for enabling users to build functional “spark” micro-apps that are integrated with AI capabilities and can connect to external data sources. Without the hassle of managing cloud resources, developers can draft an app from an initial prompt, see real-time previews, and store different versions. The apps are versatile, running smoothly on desktops, tablets, and mobile devices. GitHub intends to evolve Spark based on user feedback, making it accessible to developers and casual users alike.
Beyond these additions, GitHub has rolled out significant updates to Copilot’s capabilities within the Visual Studio Code editor, GitHub Copilot Workspace, and GitHub Models for generative AI app development. Copilot Chat now includes a multi-file editing feature, enabling simultaneous changes across files using natural language prompts—a feature available from November 1. Also new, GitHub Copilot Autofix introduces security-focused campaigns, empowering developers and security teams to address vulnerabilities across extensive codebases, managing up to 1,000 alerts in a single batch.
Looking ahead, GitHub Copilot Extensions are set to be generally available in early 2025. These extensions, currently in preview, enable Copilot users to consult integrated tools and services like Atlassian Rovo, Docker, Sentry, and Stack Overflow. Additionally, GitHub recently rolled out Copilot’s code completion capabilities in a public preview for Apple’s Xcode IDE, extending its reach to iOS and macOS developers.