A Claude Code alternative built around a crew of AI agents
Claude Code is Anthropic's coding agent, used mainly from the terminal. VibeCode.surf takes a different route: a desktop app where several agents plan, write, run and test your code, with any model you choose, on your own machine.
Reasons people look for a Claude Code alternative
You want to see it, not only read it
A live preview, a file tree, diffs and the agents' browser screenshots, beside the chat.
You want other models too
Use Claude where it shines and other providers or local models for the rest, in one project.
You want a visible crew
Each agent has its own tab, role and model, and you can talk to any of them.
What VibeCode.surf offers
A crew of agents
A lead agent splits a job across up to six helpers that work at the same time, each with its own role and model.
Agents that cannot clash
Each helper owns its files, nobody can overwrite a file another agent changed since they read it, and commands run one at a time.
Compare mode
Send one request to two models at once. Each works in its own copy of the project; you look at both results side by side and keep the better one.
Any model, your key
Claude, OpenAI, OpenRouter, Gemini, Groq, DeepSeek, Z.ai (GLM), Moonshot Kimi, Qwen, Together AI, Mistral, xAI, or local models through Ollama and LM Studio. A different model per agent.
Self-testing in a real browser
Agents open your project on localhost, take screenshots, click, type and read the console at desktop, tablet and phone size.
A plan and a memory in your repo
PLAN.md holds phases and task status; AGENTS.md is long-term memory, updated after each working turn.
Cost on screen
Tokens and cost under every reply, a total per project and per model, and each model's price in the picker.
Undo for everything
A checkpoint before and after every agent turn. Undo, redo, or restore the project to any message.
Images, voice and video
Agents generate images in the shape you ask for, record voice-overs and create video, saved into the project.
On your machine
A desktop app. Your code stays in your folder, your keys stay on your computer, and a free account keeps the app up to date.
Where Claude Code is the better choice
Different tools suit different work. Claude Code is a strong pick if you care most about:
- Working entirely in the terminal and over SSH
- Scripting, automation and continuous-integration use
- First-day access to Anthropic's newest agent features
For what Claude Code offers today, and its pricing, see its own website. Claude Code is a trademark of its owner; VibeCode.surf is not affiliated with it.
What VibeCode.surf does not have yet
It is a beta. If you depend on any of these every day, keep your current editor alongside it for now.
- Autocomplete and go-to-definition from language servers
- Inline AI edits and tab completion inside the editor
- A debugger with breakpoints
- An extension marketplace
Good to know
Can I use Claude models here?
Yes, with your own Anthropic API key or through a router such as OpenRouter.
Does it read CLAUDE.md?
Project memory here is AGENTS.md. Copy what matters from your existing file into it.
Which systems does VibeCode.surf run on?
macOS, Windows and Linux, installed with one command. The Mac version has had the most testing so far.
What does it cost?
The app is free during the beta. You pay your model provider for what the agents use, or nothing with a local model.
Install it from your terminal.
One command installs the desktop app on macOS, Windows or Linux. Free during the beta; bring your own API key or a local model.
curl -fsSL https://vibecode.surf/install.sh | shirm https://vibecode.surf/install.ps1 | iexcurl -fsSL https://vibecode.surf/install.sh | sh