Google has launched Antigravity, a development platform that transforms how AI agents write and test code by giving them direct control over editors, terminals, and browsers—transforming them from helpful assistants into independent software developers that can build, test, and verify their own work.
Google provides “generous rate limits” for Gemini 3 Pro that refresh every five hours, claiming only “a very small fraction of power users” will hit them. The platform centers on four design principles: trust through transparent reporting, autonomy via direct tool access, continuous feedback through inline annotations, and self-improvement by learning from past projects.
The platform runs on Gemini 3 Pro—which tops the LMArena leaderboard with a 1,501 Elo score and achieves 76.2% on SWE-bench Verified—and additionally supports Claude Sonnet 4.5 and GPT-OSS. It’s available for download on Windows, macOS, and Linux.