# Zed — Rust-based code editor with a built-in agent panel, ACP external agents and Zed-hosted models.

> Source: The Agents Index — https://theagentsindex.com/zed (structured, researched, re-verified)
> Facts last verified: 2026-09-29

Zed is an open-source, Rust-written code editor whose Agent Panel runs a native agent, ACP external agents such as Claude or Codex, and terminal agent CLIs. Models come from Zed's hosted service, your own API keys, or local servers.

| Fact | Value |
| --- | --- |
| Website | https://zed.dev |
| Price | from $10/month |
| Free tier | Yes |

Personal plan is $0 forever with 2,000 accepted edit predictions and unlimited use with your own API keys or external agents; Zed-hosted models need Pro. Separately, a 14-day Pro trial gives $5 of GPT-5.6 Luna usage with no credit card, and converts to free when it ends.

## What it does

**Writes code** · Yes — The Agent Panel authors and edits code from a written task description, with the agent gathering its own context and writing files through edit_file and write_file tools.

> Describe the task, the agent does the rest: discovering context, editing files, and staying on track until it's done.
> — https://zed.dev/ai · Feature card, 'Agentic Editing'

**Debugs and fixes** · Yes — Vendor names debugging as a core Agent Panel use, and the agent has a diagnostics tool that pulls language-server errors per file or project-wide to drive the fix.

> It's the core of Zed's AI code editing experience — use it for code generation, refactoring, debugging, documentation, and general questions.
> — https://zed.dev/docs/ai/agent-panel · Intro paragraph, Agent Panel page

**Reviews code** · Partly — Review is aimed at the agent's own diff, accepted or rejected hunk by hunk in a multibuffer; nothing in the docs describes reviewing a pull request another person wrote.

> Review the agent's work in an editable unified diff. Accept or reject all changes before committing them.
> — https://zed.dev/ai · Feature card, 'Review and Refine'

**Whole-codebase work** · Partly — The agent searches the repository with grep and find_path and can spawn subagents for parallel investigation, but Zed asks you to supply context explicitly and documents no repository-wide index.

> The agent can search your codebase to find relevant context, but providing it explicitly improves response quality and reduces latency.
> — https://zed.dev/docs/ai/agent-panel · Adding Context section

**Ships it** · Partly — There is no deployment feature; the agent stops at a reviewable diff, though its terminal tool can execute any shell command you permit, which is how a deploy script would run.

> Executes shell commands and returns the combined output, creating a new shell process for each invocation.
> — https://zed.dev/docs/ai/tools.md · Edit Tools section, `terminal` tool

**Done-for-you service** · No — Zed sells software you download and run yourself; the plans page offers only self-serve subscriptions, with sales contact reserved for larger order forms.

> No minimum. Business is self-serve and available to organizations of any size. Month-to-month plans are available to everyone. Order form contracts are available at 25+ seats
> — https://zed.dev/pricing · Pricing FAQ, 'Is there a minimum seat count for Business?'

_Not established: Writes tests_

## How much it does unattended

**Runs autonomously** · Partly — The agent works a task through to completion on its own, but permission-gated tool calls default to asking first, so unattended running requires setting tool_permissions.default to allow.

> "confirm" (default) — Prompts for approval before running any tool action, including MCP tool calls
> — https://zed.dev/docs/ai/mcp · Tool Permissions section, permission list

**Multi-agent** · Yes — A parent agent can delegate to subagents with their own context windows, and separate threads run concurrently, optionally isolated in their own Git worktrees.

> Spawns a subagent with its own context window to perform a delegated task. Useful for running parallel investigations, completing self-contained tasks, or performing research where only the outcome matters.
> — https://zed.dev/docs/ai/tools.md · Other Tools section, `spawn_agent`

**Agent permissions** · Yes — Permissions are per tool and pattern-based, with allow, deny and confirm rules, plus hardcoded rules for recursive deletes that no setting can override.

> Configure which Agent Panel tools run automatically and which require your approval.
> — https://zed.dev/docs/ai/tool-permissions.md · Tool Permissions page, opening line

## Which models it runs on

**Claude** · Yes — Anthropic Claude models are served through Zed's own hosted plan with published per-million-token rates, and Claude is also reachable via your own Anthropic key.

> | Claude Sonnet 5   | Anthropic | Input               | $2.00                        | $2.20                   |
> — https://zed.dev/docs/account/zed-hosted-models.md · Zed-Hosted Models pricing table, Claude Sonnet row

**GPT** · Yes — OpenAI GPT models are in the hosted catalogue with published token rates, and the Pro trial runs on a GPT model as its only hosted option.

> | GPT-6 Sol         | OpenAI    | Input               | $2.00                        | $2.20                   |
> — https://zed.dev/docs/account/zed-hosted-models.md · Zed-Hosted Models pricing table, GPT-6 Sol row

**Open models** · Yes — Open-weight models run locally through llama.cpp, Ollama or LM Studio, or any local OpenAI-compatible server; the docs walk through pulling and selecting such a model.

> Use local models when you run the model on your machine or on infrastructure you control.
> — https://zed.dev/docs/ai/use-a-local-model.md · Use a Local Model, opening line and provider table

**Model choice** · Yes — You choose the model from a selector and can route through Zed hosting, your own API keys, an existing subscription, a gateway or a local server.

> Use this page to choose which models power the Zed Agent and other Zed-owned AI features
> — https://zed.dev/docs/ai/llm-providers.md · LLM Providers, intro and 'Choose a Model Access Path' table

## Where you use it

**In your editor** · Yes — Zed is itself the editor: the agent lives in a panel beside the buffers, and edits stream into the same window you are working in.

> Smoothly collaborate with agents in Zed, editing files, navigating code, and running tools at the speed of the editor itself.
> — https://zed.dev/ai · Hero paragraph, 'AI coding at native speed'

**On the command line** · Partly — The zed CLI opens files and can be driven from scripts, but it launches the editor rather than running an agent headlessly; agent CLIs instead run as Terminal Threads inside Zed.

> Use Zed's command-line interface (CLI) to open files and directories, integrate with other tools, and control Zed from scripts.
> — https://zed.dev/docs/reference/cli.md · CLI Reference, opening line

**In your pipeline** · No — Zed's own inventory of agent paths is three, and each runs inside the editor's Agent Panel, Threads Sidebar or terminal; no pull-request or CI runner surface is offered.

> Zed supports three agent paths. Choose the path based on how you want agentic work to run.
> — https://zed.dev/docs/ai/agents.md · Agents page, intro plus 'Runs in' column of the agent path table

**In a browser** · Partly — The only web surface is the organisation dashboard for admin controls, usage and billing; coding and the agent happen in the installed desktop app, not a browser.

> Admin controls are available to owners and admins in the organization dashboard at dashboard.zed.dev.
> — https://zed.dev/docs/business/admin-controls · Accessing admin controls section

## Whose machine it runs on

**Self-hosted** · Partly — Models and edit prediction can run on hardware you control and the editor builds from source, but Zed's hosted AI, collaboration and org dashboard remain Zed-operated services.

> Connect models through Ollama and keep everything on your hardware. Nothing leaves your machine.
> — https://zed.dev/ai · Full control in full-speed section, 'Run locally' card

**Open source** · Yes — The vendor states the editor, its Zeta edit-prediction model and the ACP agent protocol are all open source and developed in public on GitHub.

> Zed, Zeta, and ACP are all open source. The editor, the model, and the protocol, built in public.
> — https://zed.dev/ai · Full control in full-speed section, 'Open source, open protocol, open model' card

## What it costs to run

**How it meters** · Yes — Two units at once: a per-seat monthly subscription, and hosted model usage metered per token at provider list price plus ten percent beyond the included credit.

> Usage of Zed's hosted models is measured on a token basis, converted to dollars at the rates listed on Zed-Hosted Models (list price from the provider, +10%).
> — https://zed.dev/docs/account/plans-and-pricing.md · Usage section

**Free tier** · Yes — The Personal plan is free indefinitely with no call required: the full editor, a capped allowance of accepted edit predictions, and unlimited use of your own keys or external agents.

> PERSONAL $0 forever The next-generation code editor. Free forever.
> — https://zed.dev/pricing · Plan cards, Personal column

**API access** · Partly — No public REST API for the product is documented; programmatic use is limited to the zed CLI from scripts and to extensions and ACP agent integrations built against Zed.

> Use Zed's command-line interface (CLI) to open files and directories, integrate with other tools, and control Zed from scripts.
> — https://zed.dev/docs/reference/cli.md · CLI Reference, opening line

**MCP server** · No — MCP runs one way only: Zed is the client that consumes external MCP servers and forwards them to agents, and the docs describe no MCP server exposing Zed to another agent.

> Zed uses the Model Context Protocol to interact with context servers.
> — https://zed.dev/docs/ai/mcp · Model Context Protocol page, opening line and Supported Features

**Bring your own key** · Yes — Bring-your-own-key is a first-class path on every plan including the free one, with keys stored in the system keychain rather than settings.json.

> We currently support Amazon Bedrock, Anthropic, GitHub Copilot, Deepseek, Google AI, LM Studio, Mistral, Ollama, OpenAI, OpenRouter, Vercel, and more.
> — https://zed.dev/pricing · FAQ, "I don't want to pay for Zed, can I use my own keys?"

## Buying it for a team

**A company can buy it** · Yes — The Business plan is bought by an organisation, is self-serve with no seat minimum, and order-form contracts open up at twenty-five seats and above.

> No minimum. Business is self-serve and available to organizations of any size. Month-to-month plans are available to everyone. Order form contracts are available at 25+ seats
> — https://zed.dev/pricing · FAQ, 'Is there a minimum seat count for Business?'

**Seat model** · Yes — Business is priced per seat per month with no floor and no published ceiling; Billing Manager is a role that does not consume a paid seat.

> BUSINESS $30 per seat, per month
> — https://zed.dev/pricing · Plan cards, Business column price cell

**Pooled budget** · Partly — Business pools AI spend under one org-wide limit and a single invoice, but seats carry no bundled credit allotment, so the shared pool is spend, not an allowance.

> Business seats don't bundle a fixed LLM credit allotment. You can use Zed-hosted AI at standard usage rates, or bring your own API keys
> — https://zed.dev/pricing · FAQ, 'Do Business seats include AI credits?'

**Admin controls** · Partly — Business admins can disable hosted models by model, turn off edit predictions and lock data sharing, but the controls do not reach BYOK, gateways, local models, external agents or extensions.

> These controls don't cover bring-your-own-key (BYOK) configurations, gateways, local or self-hosted models, External Agents, Terminal Threads, or third-party extensions
> — https://zed.dev/docs/business/admin-controls · Admin Controls, opening paragraph

**Audit log** · No — Zed's own account of Business admin capability covers AI feature availability, data egress and spend tracking; no administrative audit or activity log is documented anywhere in the Business docs.

> Administrators control which AI features are available, what data leaves your environment, and how AI spend is tracked.
> — https://zed.dev/docs/business/overview · Zed Business page, opening paragraph and What's included list

**Single sign-on** · No — Zed states plainly that enterprise identity is not there yet; members authenticate via GitHub after accepting an invitation instead.

> SSO, SAML, and SCIM are planned but not currently available.
> — https://zed.dev/pricing · FAQ, 'What admin controls are available in Business?'

## What happens to your code

**Opt out of training** · Yes — Nothing is retained or trained on unless you rate a response or opt into training collection; on Business both are off by default and members cannot opt in.

> Zed only retains AI data when you explicitly share feedback or opt in to training data collection.
> — https://zed.dev/docs/ai/privacy-and-security · AI Privacy, intro paragraph

**Data residency** · Yes — Zed's websites, software and service are hosted in the United States, with EEA/UK transfers under Standard Contractual Clauses.

> Our websites, products, and services are hosted in the United States
> — https://zed.dev/privacy-policy · Privacy Policy, International Visitors section

**Getting out** · Partly — Threads can be opened and copied out as Markdown, and monthly included credit resets rather than accumulating, but no bulk account data export is documented.

> Monthly included credit resets on your monthly billing date.
> — https://zed.dev/docs/account/plans-and-pricing.md · Usage section

**Certifications** · No — Certification is in progress rather than held: Zed reports working toward SOC2 Type 1 and points compliance questions at sales, with no ISO or GDPR attestation published.

> Zed is working toward SOC2 Type 1 certification.
> — https://zed.dev/docs/soc2.md · SOC2 page, only sentence

## Platform record (as of 2026-08-30)

Reported by the code-hosting platform, not by the vendor.

| Fact | Value |
| --- | --- |
| Stars | 89447 |
| Forks | 10332 |
| Last push | 2026-08-30 |

## Provenance

1 of 38 sources behind the facts above are independent of the vendor.

## Coverage elsewhere

Pages about this product that the vendor does not control. The label states how independent each one is.

| Source | Kind | Read |
| --- | --- | --- |
| en.wikipedia.org | Platform record | 2026-08-23 |

## What changed

- 2026-09-29 — Capabilities · Commercial terms · Free tier
- 2026-09-28 — Capabilities · Commercial terms · Free tier
- 2026-09-03 — Capabilities · Commercial terms · Free tier
- 2026-08-31 — Capabilities · Commercial terms · Free tier · Price
- 2026-08-19 — Summary

## Correcting this record

Anything marked "Not established" is a question we have not answered — it is not a No. If you represent Zed, or anything here is wrong or out of date, you can supply it with a source: the operations this site accepts from an agent are listed at https://theagentsindex.com/.well-known/agents.json — that catalogue needs no key to read, and each entry states its own auth.

Supplying evidence is free. Nothing that can be bought here changes what a field says, whether a gap is closed, or how this record is scored.
