Zed reached 1.0 yesterday — a milestone five years in the making. For an editor that’s been in heavy public beta since 2022, with hundreds of thousands of users already running it daily, the version number is symbolic rather than functional. But it’s the kind of symbolism that triggers a real wave of evaluations from teams that had Zed on the “maybe in 2026” list. Worth a careful look at what 1.0 actually means — and what the editor space will look like in six months as a result.
What 1.0 actually shipped
- Cross-platform. macOS, Linux, and Windows all GA in the same release. Windows was the long pole; the team has been polishing it for most of 2025.
- Multi-language LSP support, debugger, Git integration, SSH remoting — the standard editor surface. Nothing remarkable individually, but together it covers the “basics that prevent adoption” list that’s haunted Zed since 2022.
- AI-native architecture via the new Agent Client Protocol. Zed treats Claude Code, Codex, OpenCode, and Cursor’s agent as first-class participants in the editor, not as a chat sidebar. Multiple agents can run in parallel, with keystroke-level edit predictions.
- Zed for Business — centralized billing, role-based access, team management. The commercial half of the strategy.
The line that most matters to me is “built like a video game” — Zed has its own GPU UI framework (GPUI) and is written in Rust, not Electron. This isn’t marketing. It shows up as “the editor stays at 120fps when you scroll a 50,000-line file.” If you’ve used VS Code on a battery-constrained laptop, the difference is immediately felt.
Who this is for, who it isn’t
- Likely to switch: Sublime Text holdouts who never made the VS Code jump (Zed’s keybindings and feel are a clear descendant of Sublime + GPUI’s responsiveness). Vim/Neovim users who want a GUI with first-class Vim mode — Zed’s mode is closer to faithful than VS Code’s plugin. Anyone who works on M-series Macs with battery anxiety.
- Won’t switch: Teams deeply invested in JetBrains’ refactoring tooling, anyone using a niche language with a lone-maintainer extension only available for VS Code, anyone who has built workflows around VS Code’s Remote-Containers / dev-containers ecosystem.
- Won’t switch yet: Heavy Cursor users who depend on Cursor’s specific Composer flow. Zed’s Agent Client Protocol is interesting in theory, but Cursor’s own product still has sharper UX for the “multi-file refactor with AI” case.
The deeper trend
What’s actually happening in the editor space is the unbundling of VS Code’s monopoly. Three years ago, VS Code was the only sane mainstream choice; the only competitors were JetBrains (paid, slow, heavy) and the “handwritten config in Vim” tribe. Now there are four credible mainstream options: VS Code, Cursor (a VS Code fork with AI prepended), JetBrains, and Zed. None are dying; all are gaining.
The reason is that “editor” got bigger as a product surface. AI-assisted editing went from optional to expected. Pair-coding agents went from “tools that suggest a line” to “agents that take five steps autonomously.” Performance differences between Electron and native became user-visible because everyone has Cursor running alongside three other Electron apps and noticed the laptop fan. Each of these axes opens room for a new contender, and Zed has been positioned for the moment when those axes matter at the same time.
The Agent Client Protocol is the bit worth watching most carefully. If it’s a real standard — and Anthropic, OpenAI, and the OpenCode teams have all signed on — then editor and AI coding agent become decoupled in a way they currently aren’t. Today, choosing Cursor commits you to Cursor’s agent. Tomorrow, on Zed (or on any editor that adopts the protocol), you choose your editor and your agent independently. That’s a more competitive market for both.
What to do this week
- Install Zed and run a real project through it for a day. Don’t migrate — just spend 8 hours of actual work in it. The Vim mode, the AI integration, the perf curve under a 100k-line repo. You’ll know within a day whether your fingers can land here.
- If you maintain a VS Code extension, check whether Zed has an analog. The Zed extension ecosystem is much smaller; if your tool ships through VS Code, you might also need to ship through Zed soon to retain users.
- If you’re on a team, don’t push for a wholesale switch yet — let the early adopters report back. Zed for Business is brand new; the ergonomics of team rollouts haven’t been tested at scale.
- Watch the Agent Client Protocol. If you’re building on top of Claude Code, Codex, or any agent, the protocol will likely become the cross-editor abstraction layer. Implementing against it now is cheaper than retrofitting later.
The longer-term question Zed’s 1.0 raises: what’s the ten-year settlement of the editor market? My guess is two native-feeling editors (Zed and JetBrains), one Electron-based dominant default (VS Code, possibly with Cursor as the AI fork), and a long tail of Vim/Helix/etc. for the “I configure my own” segment. We’ve been in a one-editor-fits-all era since 2018; that era is ending, and 1.0 is one of the markers of the new shape.
Source: Zed 1.0 announcement on the official Zed blog.
Photo: React code displayed in a code editor by Antonio Batinic on Pexels.