Xode is an open-source AI dev environment that writes, runs, and deploys code in real time. Powered by your choice of LLM — Claude, GPT-4o, Gemini, Mistral, and more.
From first prompt to production. Xode handles the full stack — in-browser, no setup required.
Generate React, Next.js, Node, Python, and more from a plain English prompt. Handles routing, APIs, auth, and databases.
Your code runs in a real WebContainer — no VMs, no cloud VMs to spin up. Instant feedback on every change.
Switch between Claude Sonnet, GPT-4o, Gemini 1.5, Mistral Large, and Llama without changing your workflow.
Push to Netlify, Vercel, or Cloudflare Pages with a single click. Your app is live in under 30 seconds.
Push to GitHub, create branches, open PRs — all from inside Xode. Your workflow, not a replacement for it.
Use your own Anthropic, OpenAI, or Google keys. No middleman markup. Full control over cost and usage.
Pick the smartest model for the job. Switch mid-project without losing context.
Best for complex reasoning, long context, and architectural decisions.
Fast, multimodal. Excellent for UI generation and quick iterations.
Google's powerhouse. Massive 2M token context window.
Best open-source performance. Self-hostable with Ollama.
Type a prompt in plain English. Be as specific or vague as you like — Xode asks clarifying questions when it needs them.
Xode writes every file, installs packages, and runs your app in a real environment — all in your browser tab. No CLI, no Docker.
Chat with Xode to refine anything. When you're happy, deploy to Netlify or push to GitHub — one click, 30 seconds.
Real feedback from developers, founders, and makers.
"I shipped a full SaaS MVP in a weekend. Xode wrote the auth system, Stripe billing, and dashboard in about two hours of back-and-forth."
"Finally an AI code tool that doesn't hallucinate broken imports. The WebContainer approach is genuinely different — it actually runs the code."
"I'm not a developer and I built my client portal in a day. The multi-model switching is clutch — Claude for structure, GPT-4o for UI polish."
Bring your own API keys and the model costs come directly to you — no markup.
Forever free. BYOK required.
Includes $5 of hosted model credits.
Per seat, billed annually. Min 3 seats.
Start building for free. No credit card required.
Already have an account?
Sign in to your Xode workspace.
Don't have an account?
Xode is an open-source, browser-based AI development environment forked from bolt.diy by StackBlitz Labs. It lets you build full-stack applications using any major LLM — no local setup, no Docker, no CLI.
Xode is a fork of the open-source bolt.diy project. We've extended it with hosted model keys, team workspaces, one-click deploy integrations, and a redesigned UI. The core WebContainer runtime is the same.
When you type a prompt, Xode sends it to the selected LLM with a carefully engineered system prompt that instructs the model to return structured file operations. Those operations are applied to an in-browser file system (StackBlitz WebContainers), which runs a real Node.js runtime — including npm, package installs, and a dev server.
This means the code Xode writes actually runs in the browser tab — not in a simulated environment or a remote VM. You see real console output, real errors, and a real preview URL.
| Model | Provider | Context |
|---|---|---|
| claude-sonnet-3-7 | Anthropic | 200k |
| gpt-4o | OpenAI | 128k |
| gemini-1.5-pro | 2M | |
| mistral-large | Mistral AI | 32k |
| llama-3.1-70b | Ollama (local) | 128k |
| + 7 more | Various | — |
Choose where to deploy. Your app will be live in under 30 seconds.
Deployed successfully. Your app is now available at: