Google Antigravity announced on November 11 that browser use is rolling out in the Antigravity App: the agent can open pages in the side-panel built-in browser, interact with your running app, inspect console errors and failed network requests, take screenshots and verify fixes before finishing. Per the official docs, the built-in browser agent needs App v2.30.0+ and is rolling out on macOS, Windows and Linux to personal accounts on Free, Plus, Google AI Pro and Ultra; business accounts and the CLI still use /browser with local Chrome 144+.
Key Takeaways
- ✓Scope: App v2.30.0+, rolling out to personal accounts (Free, Plus, AI Pro, Ultra) on macOS, Windows and Linux; not yet for business accounts, Remote Agents or SSH workspaces
- ✓Parallel: several browser subagents can run at once, each owning one tab that cannot touch other subagents' tabs or yours
- ✓Model: browser subagents default to Gemini Flash in their own context window; if Flash quota runs out the agent stops and asks instead of switching models
- ✓Safety: downloads (v2.31.0+) pause for approval one by one, executables such as .exe, .sh and .dmg are blocked; uploads are capped at 20 staged files, 50 MiB each
- ✓Sign-in: the built-in browser keeps an isolated cookie profile per account; SSO, MFA and CAPTCHA pages pause for you, and payment and identity pages are off-limits to the agent
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background and pain points
A common failure of coding agents is declaring work done when the page never rendered, the form failed or the mobile layout broke, because the agent never looked. Until now, Antigravity needed you to install Chrome, enable remote debugging and type /browser. Building the browser into the desktop app puts research, coding and live verification in one loop.
Architecture
You describe what to verify in plain language. When a task needs live rendering, clicks, responsive checks or console and network inspection, the main agent spawns a browser subagent automatically; for static docs it just reads the text. The subagent acts through structured browser actions in an isolated worker with no shell or workspace tools and no arbitrary in-page JavaScript, and network logs strip cookies, Authorization headers and OAuth/JWT tokens. New sites need approval, localhost is isolated per port, and sensitive Google subdomains such as Gmail, Drive and Cloud Console need their own rules.
Benchmarks
Google has not published success rates, timings or comparisons with browser features in Cursor or Claude Code. What is confirmed is the version gating, limits and the default model (Gemini Flash) in the docs.
Getting started
Update the Antigravity App to v2.30.0+ and check Browser Actuation Rules under Settings > Browser: "no separate browser install is needed" means the rollout reached you, "Requires Google Chrome" means not yet. Then ask, for example, "start the dev server, open the signup page and check the form at 390px width". Sign in only with test accounts and preset Allow, Ask and Deny domains in the actuation rules.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.