Zed

Zed is a modern, high-performance code editor built from the ground up to feel fast, fluid, and collaboration-first: created by the team behind Atom and Tree-sitter, it’s written largely in Rust and uses GPU-accelerated rendering and multi-core architecture to make editing and navigating large codebases feel instantaneous. Unlike many Electron-based editors, it emphasizes native performance and a tightly integrated developer experience, bundling features you expect—rich syntax awareness via Tree-sitter, built-in Git workflows, an editor-first debugger, and extensibility through an extensions ecosystem—while also leaning into experimental capabilities such as multimodal “agentic” AI tools that let you converse with language models, request codebase-wide edits, and review proposed changes as editable diffs inside the editor. Collaboration is a core part of the product: Zed supports real-time multiplayer editing with chat and shared cursors so teams can work together without hopping between separate meeting or pairing tools.

The AI story is pragmatic and privacy-minded: Zed provides an Agent Panel and Edit Prediction features that integrate LLMs into the normal workflow, supports a variety of models and self-hosted options, and surfaces every agent suggestion as a reviewable change so developers remain in control. The project is open source—hosted on GitHub—so the community can inspect, contribute, or build on the editor’s foundations, and the team maintains active documentation and release notes that show frequent iteration. Platform support today centers on macOS and Linux, with a Windows build in beta and a public download channel and install scripts for supported platforms. It has also received notable venture backing to accelerate its vision: in 2025 the company announced a $32 million investment led by Sequoia to scale collaborative AI features and expand platform support. All of this combines into an editor that aims to be not just faster, but conceptually different: an environment where human developers and AI agents collaborate directly in the code, where real-time teamwork is first class, and where performance and control are prioritized so the tooling fades into the background and developers can focus on the work itself.