High-performance editor core
Zed is written from scratch in Rust and is designed to use multiple CPU cores and the GPU efficiently for a fast editing experience.
Zed est un éditeur de code multijoueur haute performance pour macOS, Linux et Windows, avec IA agentique, collaboration en temps réel et accès flexible aux modèles.
Zed is a high-performance code editor built for collaborative software development with humans and AI. It is available for macOS, Linux, and Windows, and the homepage describes it as minimal, fast, and crafted for teamwork.
The editor is written in Rust and emphasizes native-speed interaction, multiplayer collaboration, and agentic workflows. Zed Agent can read, edit, search, and run code in a project, while other model-backed features include Inline Assistant, Edit Prediction, Parallel Agents, and Git commit generation.
Zed also gives users control over how AI is used. You can rely on Zed-hosted models, bring your own provider keys or subscriptions, use gateways or local models, or disable AI entirely. Pricing is organized into Personal, Pro, and Business plans depending on whether you want no AI, individual AI access, or org-level controls.
Zed is written from scratch in Rust and is designed to use multiple CPU cores and the GPU efficiently for a fast editing experience.
The editor supports true multiplayer collaboration, letting teammates chat, code together, and share a screen and project.
Zed Agent can read, edit, search, and run code in a project, and it works with built-in tools, Agent Profiles, Skills, Instructions, and MCP servers.
You can run multiple AI tasks at once with Parallel Agents, and Zed also supports Edit Prediction, Inline Assistant, and Git commit generation as separate AI features.
Zed can connect to Zed-hosted models, provider API keys, existing subscriptions, gateways, and local or self-hosted models through its LLM provider settings.
The AI quick start notes that AI can be fully disabled, and the pricing page describes separate Personal, Pro, and Business plans for different usage and control needs.
Use Zed Agent when you want an assistant that can inspect code, make edits, search the repository, and run tools while you review changes in the agent panel.
Use multiplayer collaboration for live pair programming, teammate walkthroughs, or shared debugging sessions where chat and screen sharing matter.
Use the provider settings when your organization already has AI contracts, wants local models, or prefers to route model access through a gateway instead of Zed-hosted models.
Use the Business plan when you need org-wide controls over AI model usage, data sharing, and billing across seats.
Use the editor without AI features when you only want a fast code editor and prefer to work with external agents or no AI at all.
Use Parallel Agents, Edit Prediction, and Inline Assistant when you want different kinds of AI help for code changes, completions, and rewriting text.
Zed can be used with Zed-hosted models, your own API keys, an existing subscription, a gateway, or a local model. The setup path depends on which AI feature you want to use and where you want model access to live.
Yes. The pricing page states that the Personal plan can be used without AI capabilities, and the AI quick-start page explains that disabling AI turns off the Threads Sidebar, Agent Panel, Edit Prediction, and Inline Assistant.
Business is the org-level layer for teams that need centralized control. Admins can restrict or disable Zed-hosted AI models by model, turn off Edit Predictions org-wide, and lock data sharing settings so members cannot override them.
Zed supports a two-week Pro trial. The trial includes $20 of token credits and unlimited accepted edit predictions, and it ends when the credits are used or two weeks pass, whichever comes first.
Zed currently supports a range of provider paths, including Amazon Bedrock, Anthropic, GitHub Copilot, Deepseek, Google AI, LM Studio, Mistral, Ollama, OpenAI, OpenRouter, and Vercel, plus local models and Zed-hosted models.
Les données de trafic sont fournies à titre indicatif uniquement.
| avr. | 1379319 |
|---|---|
| mai | 1847886 |
| juin | 1447494 |
Les analyses de trafic ne sont pas encore disponibles.
hermes-agent.nousresearch.com
Agent IA autonome pour ordinateur et terminal sur macOS, Windows et Linux, avec mémoire persistante, intégrations de messagerie, automatisations planifiées et plusieurs backends de déploiement.
talkcody.com
TalkCody est un agent de programmation IA gratuit et open source pour macOS, Windows et Linux, avec plusieurs fournisseurs de modèles, un stockage local et des flux hors ligne.
browseros.com
BrowserOS est un navigateur IA open source doté d’agents locaux pour automatiser les tâches web à partir d’instructions en langage naturel. Compatible avec macOS, Windows, Linux, les extensions Chrome et les intégrations MCP.
0011.ai
0011.ai est une plateforme de développement assisté par l’IA gérée, avec un workflow visuel dans le navigateur et un assistant pour Claude Code et OpenAI Codex dans le terminal. Elle inclut le suivi de l’utilisation, la gestion des fichiers de projet et la prise en charge de macOS et Windows.
agentgrid.sh
AgentGrid est une application de bureau qui permet de gérer plusieurs agents IA sur un canevas infini, avec terminaux, navigateurs, notes et outils de code dans le même espace de travail. Elle s’adresse aux développeurs qui orchestrent visuellement des tâches parallèles d’agents sur macOS, Windows et Linux.
onlook.com
Onlook est un éditeur de code visuel pour les designers et équipes produit, avec édition React en direct, modifications assistées par IA et auto-hébergement.