GitHub Copilot Expands Ecosystem
GitHub has expanded its Copilot ecosystem, announcing full support in the Rust-based Zed code editor alongside VS Code and JetBrains. Business and Enterprise users can now select which AI models their coding agents use, optimizing for speed or cost. Additionally, the usage metrics API now includes pull request throughput and time-to-merge, allowing teams to better quantify Copilot's impact on productivity.
- The newly supported Zed code editor was created by the team behind the Atom editor and Tree-sitter parsing framework; it is written in Rust and uses GPU acceleration for faster performance. - The model picker for Business and Enterprise users includes options from both OpenAI (GPT-5.2-Codex, GPT-5.3-Codex) and Anthropic (Claude 4.5 and 4.6 variants), or an "Auto" setting that optimizes for speed and availability. - The new productivity metrics address the challenge of proving ROI for AI tools by shifting focus from user activity to production outcomes, such as the cycle time of pull requests created by Copilot's coding agent that are successfully merged. - GitHub also recently launched a Copilot SDK in technical preview for languages including TypeScript, Python, and Go, allowing developers to embed Copilot into their own custom tools and AI-driven workflows. - The expansion comes as the AI coding assistant market grows more competitive, with tools like Cursor gaining favor for their ability to handle multi-file context in large codebases, a task where Copilot can sometimes fall short. - To improve context-awareness in specific domains, Copilot in VS Code is being integrated more deeply with extensions; for C++ developers, it can now invoke tools that use the project's specific CMake build and test configurations. - The usage metrics API, however, does not capture all user activity; it relies on telemetry from IDEs and currently excludes interactions from GitHub Mobile, the Copilot CLI, and the chat interface on GitHub.com.
