Google Launches Antigravity Platform, Redefining Agentic Development with Gemini 3 Integration

Image for Google Launches Antigravity Platform, Redefining Agentic Development with Gemini 3 Integration

Google has officially introduced Antigravity, a new agentic development platform designed to revolutionize software creation by enabling autonomous AI agents. The launch was highlighted by Omar Sanseviero, a Developer Experience Lead at Google DeepMind, who announced the platform's availability via a tweet stating, "> Introducing Antigravity (@antigravity) An agentic development platform, including task lists, verifications, browser integration, and more. Try it! https://t.co/yw0jTos7Iy Blog https://t.co/Wl9HkhGeEq".

Antigravity aims to evolve the traditional Integrated Development Environment (IDE) into an "agent-first" experience, powered by Google's latest Gemini 3 model. The platform allows developers to manage and orchestrate multiple AI agents that can operate asynchronously across various workspaces, including the editor, terminal, and browser. This approach shifts the developer's role towards managing AI agents, delegating complex tasks such as background research, coding, debugging, and validation.

A core innovation of Antigravity is its focus on "artifacts," which are structured deliverables like task lists, implementation plans, screenshots, and browser recordings. These artifacts are designed to build user trust by providing verifiable evidence of the agent's work, addressing a common challenge in AI-assisted coding where transparency can be lacking. The platform also offers an "agent-first Manager surface," acting as a mission control for spawning and observing agents.

Built on the advanced reasoning and tool-use capabilities of Gemini 3, Antigravity also supports other models such as Anthropic's Sonnet 4.5 and OpenAI's open-source gpt-oss. The platform is available in public preview, offering generous rate limits on Gemini 3 Pro usage. Google's vision for Antigravity is to foster a more collaborative development environment, emphasizing trust, autonomy, feedback, and continuous self-improvement in agentic workflows.