Google Antigravity
Google Antigravity is an agent-first development platform that gives multiple AI agents direct access to an editor, terminal, and browser to plan, write, test, and verify code around Gemini and compatible models.
Google Antigravity focuses on agents that can operate developer tools autonomously while producing verifiable outputs called Artifacts. Artifacts summarize an agent’s work as implementation plans, walkthroughs, screenshots, code diffs, or browser recordings that you can inspect independently of low-level logs.
Replay one agent run below to see what each surface leaves you to review:
Antigravity now splits those surfaces across separate products: Antigravity 2.0 is a standalone desktop command center for launching, monitoring, and orchestrating multiple agents in parallel across projects and workspaces, while the separate Antigravity IDE keeps the Editor view for synchronous coding with tab completions and inline commands, similar to Cursor. Since the Antigravity 2.0 release in May 2026, it also ships an Antigravity CLI for driving the same agent harness from a terminal. That CLI supersedes Gemini CLI, which stopped serving requests for individual users on June 18, 2026, and it carries over Agent Skills, hooks, subagents, and extensions as Antigravity plugins. It supports Google models such as Gemini as well as third-party models like Claude Sonnet and GPT-OSS, and ships in four flavors for Windows, macOS, and Linux: the Antigravity 2.0 desktop app, the Antigravity IDE, the Antigravity CLI, and the Antigravity SDK, a Python framework you install with pip install google-antigravity to build your own agents.
Official website: antigravity.google
Related Resources
Tutorial
How to Use Google's Antigravity CLI for AI Code Assistance
Get started with Google's Antigravity CLI, a terminal-based AI coding agent, and use it to read, review, and refactor your Python code.
For additional information on related topics, take a look at the following resources:
