Google Antigravity
Google Antigravity is Google's agentic development platform. Its current product lineup includes a desktop workspace, terminal interface, Python SDK and IDE tools, giving developers several ways to work with coding agents. The official product page describes these surfaces separately, so choosing one starts with the environment where you want to manage your work.
Choose a development surface
Antigravity 2.0 manages multiple local agents in parallel. You can group conversations into Projects, work across multiple workspaces and automate routine tasks with scheduled messages. This is the desktop option for organizing several pieces of work without placing every conversation in the same project.
The CLI runs coding agents and shell commands from the terminal. It also manages background subagents. Developers who spend their day at a command prompt can use that interface, while the SDK provides a Python route for prototyping custom agents and running evaluations on Google's agent harness.
Work in an editor
The Antigravity IDE includes an agent manager, artifacts and codebase context. Google also offers IDE extensions for working in an existing editor. Its extension documentation lists Visual Studio Code, Visual Studio, JetBrains, Zed and Xcode.
The extensions support agent conversations in side panels, inline diffs and interactive plans. Those review surfaces give you places to inspect proposed work during a development task. Choose the dedicated IDE if you want its full environment, or check the extension guide for your current editor before changing your setup.
Check access and usage limits
The individual plan is listed at $0 per month. Google's pricing page also distinguishes Google AI Pro and Ultra access, weekly rate limits, and organization options through Google Cloud. Unlimited tab completions and command requests on the individual plan do not remove the separate agent-model rate limits.
For a first evaluation, select one surface and a bounded coding task, then inspect the changes and run your project's checks. That is a suggested way to assess fit; this directory entry does not report a hands-on performance test.
