Zed

Zed is a high-performance, multiplayer code editor from Zed Industries, Inc., the creators of Atom and Tree-sitter. It is intended for developers who want responsive local editing, language-aware tooling, Git workflows, remote development and optional AI assistance in one native desktop application.

Features

• Syntax-aware editing, language-server completions, diagnostics and navigation.
• Integrated terminal, tasks, debugger and Git panels with diff review.
• Extensions, configurable keybindings, Vim/Helix modes and themes.
• Collaborative project sessions and screen sharing on supported systems.
• Remote development, including SSH projects with environment-specific setup.
• Optional native AI agent, inline editing, edit prediction and external-agent integrations; review proposed edits and tool permissions.

Install and first use

Install brew install --cask zed, or download the appropriate official macOS DMG and move Zed.app to Applications. Open a project folder and select or install its language extension. The cask also installs the zed command-line launcher and shell completions. Stable and preview are separate channels; this listing is stable. Zed can check for updates after manual installation.

System and platform support

The official installation guide lists macOS 12 through 26 as supported and macOS 10.15/11 as partially supported: collaboration screen sharing requires macOS 12 or newer because it uses ScreenCaptureKit. Intel and Apple Silicon builds are available; use the matching architecture. The current default Apple Silicon 1.23.2 DMG is 118,562,674 bytes according to its official GitHub release asset; this is download size, not installed or project storage.

Linux builds support x86_64 and aarch64 and require a Vulkan 1.3 driver plus FileChooser, OpenURI and Secret/Secrets desktop portals. Windows supports 64-bit Windows 10 version 1903+ or Windows 11 version 22H2+, on x64 or Arm64 with DirectX 11 graphics and a current vendor driver. No official web application is currently available. FreeBSD is a source-build route, not an official downloadable package. No fixed RAM or disk minimum was confirmed in the retrieved installation guide.

Dependencies, conflicts and data

The Homebrew cask declares no additional formula dependency or explicit conflicting cask. That does not supply project compilers, runtimes, SDKs, Git/SSH setup, optional language servers, extensions or external-agent executables required by individual workflows. Configure these as needed for your project. Preview/nightly channels and other editors can share projects; review shortcut and file-association overlap rather than assuming a universal incompatibility. Linux graphics and portal dependencies differ from macOS.

Homebrew zap can remove ~/.config/zed and Zed application-support, caches, preferences, logs and saved state. Back up settings and useful conversation/project state before destructive cleanup. Hosted AI and collaboration use network services; provider data handling, credentials and permissions matter. Optional agent tools can modify files or execute commands, so review authorization and changes.

Accounts, costs and licensing

Ordinary local editing does not require purchasing an AI subscription. Collaboration and Zed-hosted AI use the relevant sign-in/service setup. Optional Zed Pro/Business hosted-model plans have model-specific usage charges and limits; do not treat them as unlimited or as necessary for basic editing. Bring-your-own API access is billed separately by the provider; supported subscriptions, gateways and local models provide alternative access paths. External agents generally manage access in their own processes. API keys saved through Zed use the system keychain; non-empty provider environment variables take precedence. Check current plan terms before subscribing.

The source is primarily GPL-3.0-or-later, with Apache-2.0 components where marked and separate third-party notices. This open-source editor license does not include free entitlement to commercial hosted AI services.

Official references

• Installation and platform requirements
• Project and licensing
• AI features and access routes
• API access and key storage
• Hosted models
• Current plans