AI
Google brings Antigravity AI agent into five IDEs with new extensions
Google's AI coding agent is now accessible through Visual Studio Code, Visual Studio, JetBrains IDEs, and Zed—allowing developers to use the tool without leaving their editor.
The New Stack
Google has expanded its Antigravity AI coding agent beyond its own desktop application, rolling out extensions that embed the tool directly into the development environments developers already use daily. The move lets engineers delegate complex coding tasks while staying in their preferred editor, rather than switching between applications.
IDE integration across platforms#
The Antigravity extensions are now available for Visual Studio Code on macOS, Linux, and Windows through the official extension marketplace. Support for Visual Studio 2026 is in preview, while JetBrains IDEs (including IntelliJ IDEA, PyCharm, WebStorm, GoLand, CLion, and Rider) as of version 2026.2.1 and Zed also gain native support. Developers can view agent conversations in a sidebar, compare inline code changes, review task plans, and orchestrate multi-step work—all without opening the separate Antigravity 2.0 desktop app.
A single Antigravity account syncs across all environments, eliminating the need for separate logins or license management in each editor.
Enterprise rollout with spending controls#
Google is making Antigravity available to organizations through Gemini Enterprise subscriptions at Standard, Plus, and Standard Emerging Market tiers. Administrators can set monthly spending caps per Google Cloud project, with usage either halting or shifting to pay-as-you-go when quotas deplete.
However, the current model pools all credits by project, region, and subscription tier rather than allocating budgets per developer or team. Individual and team-level controls are planned for later in 2026. The platform meters allowances using a rolling seven-day window, and unused credits expire when that window resets.
Token costs escalate quickly#
An average substantial coding task can burn through 150,000 to 200,000 tokens, and multi-agent handoffs introduce additional input tokens as work transfers between agents. This means teams can exhaust their monthly allocation rapidly. Microsoft and other vendors have already encountered this challenge—Microsoft introduced token budgets for its engineers after discovering some were spending hundreds to thousands of dollars monthly, while Uber reportedly consumed its entire 2026 AI coding budget within four months.
