Zed

Zed is an open-source AI code editor for developers who want agentic coding inside a fast editor. It pairs real-time agent visibility, editable diffs, and hosted, local, or bring-your-own-key AI choice.

Key Highlights

• Delegate coding tasks to an agent that can gather context, edit files, and run tools.
• Follow agent work in real time through the same multiplayer layer Zed uses for human collaboration.
• Review and refine changes in an editable unified diff before committing.
• Bring Claude Agent, Codex, or OpenCode through ACP, and extend context with MCP servers.
• Use Zeta edit predictions, an open-weight and open-data model trained on developer edits.

What Makes It Different

Zed's main difference is that agent work is visible inside the editor while it happens. The same multiplayer system used for human collaboration shows where the agent is navigating and editing, so review starts before a final patch appears.

It also avoids locking the workflow to one AI backend. The editor, Zeta model, and ACP protocol are open source.

Features & Capabilities

In everyday use, the agent can discover context, edit files, navigate the codebase, and run tools. Changes land in an editable unified diff, so you can refine, accept, or reject work before committing.

AI setup can use Zed-hosted models, local Ollama, or provider keys, with support for Amazon Bedrock, Anthropic, GitHub Copilot, OpenAI, OpenRouter, Vercel, and more. Fine-grain tool controls let teams configure what an agent can access.

User Ratings and Testimonials

Zed does not publish a public average review score. Developers using Zed praise its speed, Vim mode, context building, agent review, and use with Amazon Bedrock.

The practical limits are around paid AI and team controls. Personal does not include Zed-hosted AI, Business has no free trial, Business seats do not bundle fixed LLM credits, and SSO, SAML, and SCIM are planned rather than available.

Pricing & Value

• Personal: $0 forever, with 2,000 accepted edit predictions and unlimited use with your API keys or external agents like Claude Agent and Codex CLI.
• Pro: $10 per month, with Zed-hosted AI, unlimited edit predictions, $5 of tokens included, and usage-based billing beyond that credit. A two-week trial includes $20 of token credits.
• Business: $30 per seat per month, with org-wide AI model policies, data governance controls, unified spend visibility, role-based access controls, and unlimited edit predictions.

Personal works for editor-first, local, bring-your-own-key, or external-agent use. Pro is the paid step for hosted AI, while Business is for central policy and billing.