Zed
by Zed Industries, Inc.
A Rust-native, GPU-accelerated code editor where humans and any coding agent share one workspace
Zed is a code editor written from scratch in Rust by the creators of Atom and Tree-sitter, rendered on the GPU for very low input latency. It combines native multiplayer editing with an agent panel, open-weight edit prediction and the Agent Client Protocol, so external coding agents run inside the editor rather than replacing it.
Zed is a code editor written from scratch in Rust by Zed Industries — the team behind Atom, Electron and Tree-sitter — and it reached a 1.0 stable release on 29 April 2026 after roughly five years in preview. Its differentiator is architectural rather than featural: a custom UI framework called GPUI renders the editor through the graphics pipeline instead of a browser engine, and the core is built to spread work across CPU cores, which is why reviewers routinely measure cold start and input latency far below Electron-based editors. Zed ships natively on macOS, Linux and Windows, with the Windows build reaching parity via a dedicated DirectX 11 rendering backend, and 1.0 landed with an integrated debugger, SSH remote development and Git support included rather than delegated to extensions. On the AI side Zed provides an agent panel for delegating multi-file work, an inline assistant for transforming a selection with a language model, and Zeta2, an open edit-prediction model announced in March 2026 that feeds language-server context into the model and improved acceptance rates roughly 30% over the first-generation Zeta. Critically, Zed is not locked to its own hosted models: it implements the Agent Client Protocol, so external agents including Claude Agent, Codex CLI, Gemini CLI and OpenCode run inside Zed's interface, and it supports Model Context Protocol servers. Multiplayer editing, chat, screen sharing and project sharing are built into the product rather than bolted on. Zed Industries raised a $32M Series B led by Sequoia Capital in August 2025, taking total funding past $42M, and is building DeltaDB, a CRDT-based operation-level version-control layer intended to give humans and agents a shared, character-level view of a codebase in real time.
A staff or senior engineer on a Rust, Go, TypeScript or Python team who is latency-sensitive, wants agentic editing without vendor lock-in on models, and can live without the long tail of VS Code extensions.
A genuinely fast, GPU-rendered editor where your choice of coding agent — Zed's own, Claude Agent, Codex CLI or Gemini CLI — runs in the same window as multiplayer editing and the debugger.
At a Glance
- Category
- Developer Tools
- Pricing
- Freemium, Subscription, Usage-based
- Target Market
- Enterprise Developers, Platform Engineering Teams, CTOs, Data Scientists
- Deployment
- Open-source, Hybrid
- Founded
- 2021
- Team Size
- 11-50
Key Features
- ✓GPUI GPU-accelerated rendering
A from-scratch Rust UI framework that draws the editor through the graphics pipeline, which is what produces the measured latency advantage.
- ✓Agent panel with Agent Client Protocol
Delegate multi-file edits to Zed's agent or run external agents such as Claude Agent, Codex CLI, Gemini CLI and OpenCode in the same interface.
- ✓Zeta2 edit prediction
An open edit-prediction model announced March 2026 that incorporates language-server context and improved acceptance rate about 30% over Zeta v1.
- ✓Native multiplayer collaboration
Real-time shared editing, chat, screen sharing and project sharing built into the editor rather than added through a third-party extension.
- ✓Integrated debugger and SSH remote development
Shipped in the 1.0 release across all three platforms, with the UI local while the codebase runs on a remote server.
- ✓Bring-your-own-model and MCP support
Use your own provider API keys or external agents on any plan, with Model Context Protocol servers for tool access.
- ✓Org-wide AI governance on Business
Prompt sharing and edit-prediction training are off by default at org level and individual engineers cannot override those settings.
Capabilities
Use Cases
- •Latency-sensitive daily driving
Replace an Electron editor on large repositories where input lag and cold-start time have become a measurable daily tax on engineers.
- •Agent-assisted refactors without lock-in
Run Claude Agent or Codex CLI against a multi-file refactor inside the editor, reviewing every diff before it lands in the working tree.
- •Remote development over SSH
Keep the UI local while the codebase, language servers and debugger run on a remote build machine or cloud workstation.
- •Live pair programming and review
Two or more engineers edit the same buffers in real time with chat and screen share, without scheduling a separate call tool.
- •Controlled AI rollout for a team
Business admins disable Zed-hosted AI features org-wide, enforce model policy and track hosted AI spend per member from one place.
Ideal For
Best For
- ✓Latency-sensitive developers who find Electron-based editors sluggish on large files and long sessions
- ✓Teams that want agentic editing without committing to one vendor's model — Agent Client Protocol runs third-party agents inside the editor
- ✓Rust, Go, TypeScript and Python work, where the built-in language support and debugger are strongest
- ✓Pair programming and code review over native multiplayer editing, chat and screen sharing rather than a bolted-on extension
- ✓Engineers who want to bring their own API keys and keep AI spend on their existing provider accounts
Not Ideal For
- ✗Teams dependent on the VS Code extension long tail — Zed's marketplace is roughly 800 extensions against VS Code's tens of thousands, with no equivalent for database explorers, REST clients or AWS/Azure panels
- ✗Regulated enterprises with a compliance checklist: Zed's own Business page states plainly that SSO, SAML, SCIM and SOC 2 are not currently available and are still being worked on
- ✗Python-heavy shops that have hit the slow language-server performance multiple reviewers report, or Vim users who need complete keybinding parity
- ✗Organisations wanting codebase-wide AI indexing of the kind dedicated agentic IDEs offer — Zed does not provide it
Deployment
Market Analysis
Pros
- ✓Performance is the consistent finding across every independent review: reviewers report cold start and input latency an order of magnitude better than Electron editors
- ✓Open source under GPL-3.0-or-later with 89.3k GitHub stars and a large Rust contributor base, so the editor survives independently of the company's commercial plans
- ✓Agent-agnostic by design — the Agent Client Protocol runs competitors' agents inside Zed, which is unusual among AI editors
- ✓1.0 shipped debugger, SSH remoting and Git across macOS, Linux and Windows, closing the gaps that made it a secondary editor for years
- ✓Product Hunt reviewers rate it 5.0/5 across 46 reviews, praising the minimal interface and the absence of Electron bloat
Cons
- ✗Extension ecosystem is roughly 800 packages against VS Code's tens of thousands; database explorers, REST clients, cloud service panels and advanced Docker management have no Zed equivalent
- ✗Zed's own Business page states SSO, SAML, SCIM and SOC 2 are not currently available — a hard blocker for procurement at most regulated enterprises
- ✗Multiple reviewers report slow Python language-server performance and incomplete Vim keybindings, with at least one returning to another editor for Python work
- ✗No codebase-wide AI indexing, so agentic performance on very large repositories lags dedicated AI-first IDEs
- ✗2.5k open GitHub issues and an ongoing graphics-layer migration from blade to wgpu indicate the platform is still in motion after 1.0
Pricing
Personal
$0
- ✓Free forever
- ✓2,000 accepted edit predictions
- ✓Unlimited use with your own API keys or external agents
- ✓Multiplayer collaboration
Pro
From $10/mo
- ✓Unlimited edit predictions
- ✓$5 of tokens included monthly
- ✓Usage-based billing beyond that at API list price plus 10%
- ✓Two-week trial with $20 in token credits
Business
From $30/seat/mo
- ✓Org-wide AI model policies
- ✓Data governance controls
- ✓Role-based access controls
- ✓Unified spend visibility
- ✓Unlimited edit predictions
- ✓No minimum seats
List pricing is published in full. The editor itself is free forever and open source under GPL-3.0-or-later, so the paid tiers buy hosted AI, not the tool: Personal caps at 2,000 accepted edit predictions but is unlimited if you bring your own API keys or run an external agent. Pro is $10/month with only $5 of tokens included and everything beyond billed at provider list price plus a 10% margin, so heavy agent users should expect the effective cost to float well above the headline. Business is $30/seat/month with no minimum, carries no fixed AI credit bundle at all, and exists mainly for org-wide policy, RBAC and spend visibility. Spending caps are available, and pointing Zed at your own provider accounts removes the markup entirely.
Security & Compliance
Connect
Sources
This page was written from 7 sources, 3 on domains other than zed.dev.
- 1.zed.dev — zed.devvendor
- 2.zed.dev — pricingvendor
- 3.zed.dev — businessvendor
- 4.zed.dev — sequoia backs zedvendor
- 5.github.com — zed
- 6.producthunt.com — reviews
- 7.hn.algolia.com — search
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
CodeRabbit
AI code review and agentic change management for teams shipping human- and machine-written code
Qodo
Agentic code review and governance layer for teams shipping AI-generated code at scale
Sonar Vortex
Guardrails inside the coding agent's loop, not at the pull request
Slack Code
Agentic coding in shared Slack channels, where the whole team can see the diff