Readers like you help support this site. When you make a purchase using links on our site, we may earn an affiliate commission. All opinions remain our own.

Google Antigravity is a brand new coding app from Google, and it is completely free to download right now. Unlike traditional code editors where you type everything yourself, Antigravity lets you tell an AI agent what you want built, and the AI does the heavy lifting for you. It writes the code, tests it, and even checks if the final result looks right in a browser.
Think of it like this: instead of writing every single line of code by hand, you become the project manager.
You describe what you need (“build me a login page” or “fix the layout on mobile”) and the AI handles the rest. You can even run multiple AI agents at the same time, each working on a different task.
It is built on top of the same foundation as VS Code, so if you have used VS Code before, the interface will feel familiar.
But under the hood, it is powered by Google’s latest Gemini Pro AI model, which is what makes the magic happen.
This post covers everything you need to know: the main features, how to install it on Windows 11/10, and how to get past a few common setup issues.
Key Takeaways
- AI Does the Work: Instead of typing code yourself, you tell AI agents what to build and they write, test, and fix code for you automatically.
- It Can See Your App: The built-in browser lets agents check if your website actually looks right, not just if the code runs.
- Pick Your AI Brain: It comes with Google’s latest Gemini Pro, but you can also switch to Claude, GPT, or Grok models whenever you want.
What Is Google Antigravity?
Google Antigravity is the new “agentic development platform” from Google, available now as a Public Preview. While it looks like standard VS Code on the surface, it replaces the traditional file explorer workflow with a Mission Control interface designed exclusively for the AI era.
The groundbreaking shift here is the Agent Manager View. Unlike Visual Studio 2026, which heavily focuses on AI copilots assisting you as you type, Antigravity is built from the ground up for autonomous agents.
You assign a task like “refactor the login flow and test it in the browser” and the agent spins itself up, edits multiple files, runs terminal commands, and actually verifies its work using a built-in Chrome instance. It runs on the latest stable Gemini Pro model (running 3.1 as of this writing, but constantly updating), utilizing a massive context window to instantly understand your entire repository, external documentation, and even live screenshots of your app.
Key Features: What Can You Do?
Here are the biggest features that set Antigravity apart from a regular code editor.
The Manager View
This is the standout feature. Instead of just a chat box, you get a full dashboard where you can start multiple AI agents at once. One agent can be fixing a styling problem while another writes tests in the background. You watch them all work at the same time.
Browser Orchestration
Antigravity does not just write code; it can actually “see” your app running. Using a built-in Chrome browser, the AI agent can click buttons, fill out forms, and take screenshots to make sure the changes it made actually look correct.
“Vibe Coding”
You do not need to spell out every technical detail. Just describe what you want in plain language (“make the landing page feel more modern and clean”) and the AI figures out the code to make it happen. Google calls this “vibe coding.”
Model Flexibility
You are not stuck with just Google’s AI. Antigravity lets you swap in other AI models from companies like Anthropic (Claude), OpenAI (GPT/Codex), and xAI (Grok) whenever you want. New models come out all the time, and you can grab them from the built-in marketplace.
How to Download and Install Antigravity on Windows
Installing Antigravity is straightforward on Windows, though the launch servers occasionally still throw a few hiccups regarding authentication.
- Step 1: Download the installer for your OS directly from the official Google Antigravity downloads page.
- Step 2: Run the downloaded installer (
Antigravity_Setup_x64.exe). - Step 3: Important Note on Setup: When you first launch the app, you will be asked to sign in with your Google account to connect to the AI features.
- Step 4: If you get stuck spinning on “Setting up your account”: Change your default Windows browser to Chrome temporarily before signing in. Many users report that Edge or Brave can occasionally cause the auth token to hang. Close the app completely and try again.
- Step 5: Once successfully logged in, choose to import your existing keybindings and extensions from VS Code to hit the ground running!
Pro-Tip for Terminal Users on Windows
Sometimes the AI agent tries to run Linux commands instead of Windows commands. To fix this, go into your Custom Instructions settings and add: “Always use PowerShell Core for terminal execution.”
Google Antigravity vs The Competition
Is this the highly-anticipated Cursor killer? Here is how Antigravity stacks up against the other heavy hitters in the modern AI IDE space.
| IDE | Core Philosophy | AI Engine | Autonomy Level |
|---|---|---|---|
| Google Antigravity | Agent-First (Manager View) | Latest Gemini Pro | High (Browser/Terminal Access) |
| Cursor | Editor-First (Chat Focus) | Claude/GPT | Medium (Composer) |
| Visual Studio 2026 | Enterprise Stability | GitHub Copilot | Low (Assisted Typing) |
| Kiro IDE | Cloud-Native Environment | AWS Bedrock | Medium |
Quick Start Guide: First Steps
Ready to try it out? Here is how to get started once you have it installed:
- Open Mission Control: Don’t just open a file to start typing. Hit
Cmd/Ctrl + Ito open the Agent interface. - Spawn an Agent: Type “Scan my entire repo and create a README.md explaining the architecture.” Watch it independently read your files and generate the artifact.
- Visual Debugging: Run your localhost server. Click the “Eye” icon in the chat to let the agent actively see your browser window. Tell it: “Fix the alignment on that navbar,” and watch it visually inspect the element and apply the code fix in real-time.
Visual Tour: What Does Antigravity Look Like?
Check out some of my screenshots when I first explored the IDE interface upon its launch release on my PC.
Google Antigravity Home Screen
This is what you will see in this build when you load it up first after completing the Google account setup process.
Antigravity AI Models Selection
Here is where you pick which AI model you want to use. Gemini Pro is the default, but you have other options right in the dropdown.
The Mission Control Agent Manager
This is the Agent Manager where you can set up and run multiple AI agents at once and control how they work together.
Choosing Your Programming Language
I started with PHP, my favorite language! But as you can see, the environment supports a vast array of modern tech stacks.
Embedding the Chrome Extension
When you first prompt a visual task, Antigravity prompts you to install the crucial Chrome extension so it can “see” your browser.
Active Agent Interface (Project Obito)
This is what it looks like once you finish setting up everything and you are ready to vibe-code. Mine is called ‘Project Obito’ in honor of my favorite anime character!
How to Change Your AI Intelligence Engine
One of Antigravity’s biggest strengths is that it does not lock you into Google’s ecosystem. Because model capabilities leapfrog each other constantly (Gemini today, Claude tomorrow, GPT the next), you can easily swap the underlying brain of your autonomous agents.
Here is how to switch models safely and efficiently to ensure you are always using the smartest tech available:
- Open the Command Palette: Press
Ctrl + Shift + Pand typeModel Switcher. - Select Local vs Cloud: You can choose the official cloud provider API, or route through a local instance if you have the hardware to run open-source weights.
- Download the Associated Extension: For models not built natively into the app, you need to grab their official connectors from the Marketplace:
- Enter Your API Key: Once the extension is active, click the “Model” dropdown button in the top right of your Mission Control manager panel and paste in your respective API key (from Anthropic, OpenAI, or xAI).
Which model should you choose?
Use Gemini Pro for multi-modal tasks (screenshot analysis and massive context windows). Switch to the latest iteration of Claude Opus when you need deep, complex code refactoring and logical reasoning. Keep an eye out for Grok extensions for uncensored parsing.
⚠️ Known Issue: The “Pro Quota” Lockout Bug
As of early 2026, many Google AI Pro subscribers are experiencing a severe bug where the advertised 5-hour quota reset is failing. Instead of refreshing, users are being locked out of advanced models (like Gemini Pro and Claude Opus) for 3 to 7 days at a time.
Before relying on Antigravity for heavy daily work, we highly recommend reading the ongoing Google AI Developer forum thread to track when Google officially deploys a fix.
Ready to try Agentic Coding?
Skip the boilerplate and start managing autonomous AI software agents today.
Frequently Asked Questions
Is Google Antigravity free to download?
Currently, the Public Preview is completely free for individuals. There will likely be a paid “Pro” tier introduced later for advanced compute usage and enterprise teams.
Does it replace VS Code?
It is actually a fork of VS Code, meaning it effortlessly supports the same extensions and hotkeys. However, the workflow is fundamentally different—you are meant to manage agents, not manually type every line of code.
Is my code private?
Google states that on the Enterprise plan, code is not used for model training. For the free preview, you should carefully check the “Data & Privacy” settings during setup, where you can opt-out of training.
Can I use it for C# and .NET?
Yes, via the C# Dev Kit extension. However, for serious .NET work at a large company, Visual Studio 2026 is still the best choice in the Microsoft world.
Discover more from Windows Mode
Subscribe to get the latest posts sent to your email.






Amazing and ironically this has forced all the major Ai’s to release their next versions way faster, including Grok 4.1 is now out and even ChatGPT 5.1 🤣
Gemini 3 Pro is something else. I think this will force the other guys to step up their game for the next release.
I agree, this release was not expected and I think it’s exactly what the AI industry needed. Google is not playing around.
better than cursor or vscode!