Google · Tools

Google Antigravity

Visit website

At a glance

Price
Pricing on request
Vendor
Google

Specifications & properties

Key decision factors

Pricing model
Freemium 1
Tool form
  • Standalone IDE
  • CLI agent
1
Autonomy level
Autonomous agent 1
Model choice
Multiple models 1

Pricing

Free tier
Yes 1

Model

Available models
Gemini 3.5 Flash, Gemini 3.1 Pro, Gemini 3 Flash, Claude Sonnet & Opus 4.6, gpt-oss-120b 1

Capabilities

Repo-wide context
Yes 1
Terminal/command execution
Yes 1
Report data / suggest a correction

Google's agentic development platform with an agent-first IDE, terminal-first CLI, and SDK for running autonomous coding agents.

Profile

Google Antigravity is Google's agentic development platform — an editor, terminal, and browser combined into one workspace where you spawn, orchestrate, and observe AI coding agents rather than just chatting with one in a sidebar, built by Google and launched in public preview on November 20, 2025.

Who builds it

Antigravity is a Google product, introduced alongside the Gemini 3 model family. It ships as a downloadable desktop application for macOS, Windows, and Linux, and in 2026 expanded from a single IDE into a family of surfaces — the IDE, an "Antigravity 2.0" desktop app for higher-level task delegation, a CLI, and an SDK — that share one underlying agent harness.

Core features

  • Editor View: a familiar AI-powered code editor with tab completions and inline commands for direct, hands-on coding.
  • Manager Surface: a dedicated interface for spawning multiple agents at once and letting them work asynchronously across different workspaces, instead of one blocking chat thread.
  • Artifacts: agents produce verifiable deliverables — task lists, implementation plans, screenshots, browser recordings — so you can check what an agent actually did instead of just reading a log.
  • Browser control: agents can navigate the web, fill forms, and validate end-to-end flows without leaving the IDE.
  • Persistent knowledge base: agents retain context across sessions so future tasks can build on prior work.
  • Model choice: access to Gemini 3 Pro and Flash, Anthropic's Claude Sonnet and Opus, and OpenAI's gpt-oss-120b from within the same platform.

Pricing

  • Individual (Free): $0/month — access to the supported model lineup, unlimited tab completions and command requests, and basic weekly rate limits; Google has described this as available "at no cost for individuals" during public preview.
  • Google AI Pro tier: everything in the free plan plus more generous rate limits and a flexible AI credit pool; Antigravity's own pricing page does not list a standalone dollar price for this tier, since it is unlocked through a separate Google AI Pro subscription.
  • Google AI Ultra tier: everything in Pro plus higher access to the latest Gemini models and further-expanded rate limits, again unlocked via a Google AI Ultra subscription rather than an Antigravity-specific price.
  • Organization plan: available through Google Cloud under the Gemini Enterprise Agent Platform, billed with consumption-based API pricing rather than a flat seat fee.

Because the paid tiers ride on top of Google's general-purpose AI subscriptions (which also unlock Gemini app features outside of coding), check Google's current subscription pricing before budgeting, as the underlying consumer plan prices are set independently of Antigravity.

Who it's for

Developers and teams who want to run several coding agents in parallel on real, verifiable tasks — refactors, test generation, browser-based QA — rather than supervising one assistant turn by turn. The free Individual tier is enough to evaluate the workflow solo; the Google AI Pro/Ultra tiers suit developers who already lean on Gemini day to day and want higher throughput, while the Organization plan targets companies standardizing agentic development on Google Cloud.

Frequently asked questions

What is Google Antigravity?

Google Antigravity is an agentic development platform that combines a code editor, terminal, and browser control into one workspace so you can spawn and orchestrate multiple AI coding agents instead of using a single chat assistant. It launched in public preview on November 20, 2025 and runs on macOS, Windows, and Linux.

Is Google Antigravity free?

Yes, the Individual plan is $0/month and includes access to the supported models, unlimited tab completions and command requests, and basic weekly rate limits. Higher-throughput access requires a separate Google AI Pro or Google AI Ultra subscription layered on top of Antigravity.

What does Google Antigravity cost beyond the free tier?

Antigravity's own pricing page does not publish a standalone dollar price for its Pro and Ultra tiers; instead they are unlocked by subscribing to Google AI Pro or Google AI Ultra, Google's general-purpose AI subscriptions that also apply outside of coding. On Google's subscription page (EUR pricing at time of check), Google AI Pro starts at €21.99/month and Google AI Ultra starts at €99.99/month, with prices varying by market and subject to change.

Which AI models can I use inside Google Antigravity?

Antigravity gives you a choice of models from within the same platform: Gemini 3 Pro and Flash, Anthropic's Claude Sonnet and Opus, and OpenAI's gpt-oss-120b. Which models and rate limits you get depend on whether you're on the free Individual plan or a Google AI Pro/Ultra tier.

What platforms does Google Antigravity run on?

Antigravity is a downloadable desktop application for macOS, Windows, and Linux. In 2026 it expanded beyond the original IDE into a broader family of surfaces — an Antigravity 2.0 desktop app, a CLI, and an SDK — that share the same underlying agent system.

What makes Google Antigravity different from a typical AI coding assistant?

Instead of a single chat panel next to your code, Antigravity has a dedicated Manager Surface for spawning and running multiple agents asynchronously across different workspaces, plus a browser-control capability so agents can navigate the web and validate flows end to end. Agents also produce "Artifacts" — task lists, plans, screenshots, recordings — so you can verify what they did rather than reading raw logs.