Google has announced the launch of Antigravity, a new platform designed to empower developer teams with autonomous coding agents capable of handling complex tasks independently. Antigravity, powered by Gemini 3, represents a significant shift towards an agent-first approach, enabling agents to move beyond remote control to true autonomy.
Antigravity offers an agentic coding environment that prioritizes browser control capabilities, asynchronous interaction patterns, and an agent-first design philosophy. As the volume of code continues to surge, especially with the emergence of AI-generated code, enterprises are increasingly relying on asynchronous coding agents to streamline project reviews, evaluate components, and execute tasks autonomously.
During the public preview phase, Antigravity users can leverage Gemini 3, Anthropic’s Sonnet 4.5 models, and OpenAI’s gpt-oss to build agents compatible with major operating systems like macOS, Linux, and Windows. Google aims to position Antigravity as a cornerstone of software development in the age of agents, emphasizing trust, autonomy, feedback, and self-improvement as its core tenets.
Antigravity’s innovative approach to coding aligns with Google’s broader efforts in the coding agent space, complementing existing platforms like Jules, Gemini CLI, and Gemini Code Assist. While facing competition from other coding agent platforms, such as Codex, Claude Code, and Cursor, Antigravity’s unique features aim to enhance collaborative development environments and elevate the efficiency of coding workflows.
Early user feedback has highlighted both the potential and challenges of Antigravity, with some users reporting issues like errors and slow code generation. Despite these initial hurdles, Google’s foray into agent-first architecture signifies a significant step towards reshaping coding practices and promoting autonomous agent capabilities in software development.
Source: VentureBeat