Google Antigravity is an AI-native IDE that launched in November 2025, forked from VS Code on the surface but built around a completely different interaction model underneath. Instead of a chat sidebar suggesting one line at a time, it dispatches autonomous agents that plan a task, write code, run terminal commands, and open a built-in Chrome instance to visually check that the UI actually works.
Its two-view structure separates hands-on coding, the Editor View, from orchestration, a Mission Control-style Manager View where up to five agents can run in parallel: one refactoring a backend, another building a component, a third running tests. Each agent reports back with task lists, implementation plans, and screenshots so the results are auditable rather than a black box. It supports Gemini 3.1 Pro, Gemini 3 Flash, Claude Sonnet 4.6 and Opus 4.6, and GPT-OSS 120B, and includes a self-improving knowledge base that picks up a team's coding conventions over time.
Pricing runs from a rate-limited free tier through Pro at \(20 a month, bundled with Google AI Pro, to Ultra at \)249.99 a month for heavy users. Developers have reported tightening quotas and an opaque credit system since the initial preview, but free access to models like Claude Opus has still driven fast adoption, reportedly reaching 6% of developers worldwide within months of launch.