Google Antigravity is Google's agent development platform. Individuals are listed at $0 a month, and the Pro and Ultra cards name quotas rather than a price.

Four builds, four version numbers

The download page lists Antigravity 2.0 at v2.19.1. macOS needs 12 (Monterey) or later, on Apple Silicon or Intel. Windows is 10 (64-bit) or later, with x64 and ARM64 builds. Linux needs glibc 2.28 and glibcxx 3.4.25 or newer. The examples named there are Ubuntu 20+, Debian 10+, Fedora 36+, and RHEL 8+. Googlebook is a fourth column: Googlebook OS, pre-installed on Googlebooks, updated or reinstalled through Google Play.

The CLI on that same page is v1.2.14. macOS, Linux, and Googlebook share one installer, curl -fsSL https://antigravity.google/cli/install.sh | bash. Windows has a PowerShell script and a CMD script on the same host. The standalone Antigravity IDE is v2.5.5. The Python SDK is v0.1.18, and the page says it is co-optimized with the Gemini family. Extensions are listed for VS Code, Visual Studio 2026, JetBrains, Zed, and Xcode. The JetBrains, Zed, and Xcode lines say enterprise support is in preview.

The homepage describes 2.0 as the place to run several local agents at once, group chats into projects, work across workspaces, and send scheduled messages. The CLI is the terminal surface. The IDE is the one with the agent manager, artifacts, and the codebase. The SDK is for short Python scripts that try an agent, automate an engineering task, or run an evaluation.

A $0 card, and two cards with no dollar amount

The pricing page says Generally Available. The individual card is $0 a month, worded as Antigravity without a subscription plan. It includes a variety of agent models, unlimited Tab completions, unlimited Command requests, and basic weekly rate limits. That weekly line carries an asterisk. The page text does not spell the limit out as a number.

The Google AI Pro card and the Google AI Ultra card do not print a dollar amount. Pro says everything in the individual plan, plus more rate limits and a flexible AI credit pool. Ultra says higher access to the latest Gemini models, then everything in Pro, plus even more rate limits and the same credit pool. Get Pro points at /g1-upgrade. That page's title is "Redirecting to Google One AI..." and it does not print a Pro price either.

Organizations are a Business plan through Google Cloud. The page says Cloud customers can use Antigravity 2.0, the CLI, and the IDE extensions. API access is described as consumption-based, with no unit price on the card. Eligible Gemini Enterprise customers are shown two lines: Standard and Plus plans starting at $30 USD per seat per month, and a pay-as-you-go plan with a $0 seat fee plus consumption pricing. On the homepage, the developer block says available at no charge and links to download. The organization block says now available.

How the quota refreshes

The plans and AI credits doc says individual accounts sit under terms derived from Google's Terms of Service, and teams sit under Google Cloud terms inside Gemini Enterprise. Rate limits and model availability follow the Google AI plan. It says every plan gets Gemini models, and it names Gemini 3.1 Pro, Gemini 3.8 Flash, and other offered Gemini Enterprise models, plus unlimited Tab completions, scheduled tasks, and the CLI.

No Google AI Pro or Ultra plan: a meaningful quota, refreshed weekly, and a weekly rate limit.
Google AI Pro: a high quota, refreshed every five hours until the weekly limit is reached, and a higher weekly rate limit.
Google AI Ultra: the highest quota, refreshed every five hours, the highest weekly rate limits, and access to third-party models.

The same page says these usage limits can change, to manage capacity and keep the service stable. On Pro or Ultra, purchased AI credits, or a one-time promotional credit, can cover use past the baseline quota. The doc says those credits are consumed at standard Gemini Enterprise consumption pricing, and it does not print that unit price. The AI Credit Overages setting has two choices. Never waits until the baseline quota refreshes. Always spends credits when the baseline for that model is used up, and switches back when the baseline returns. Baseline usage across models is on the Settings page.

The doc also says two things are unsupported right now: bring-your-own-key or bring-your-own-endpoint for extra rate limits, and organizational tiers sold through a custom contract. The pricing page, on its own, still offers the Business plan through Google Cloud, including the $30 seat line and the $0 seat fee.

Four builds, four version numbers