Google has launched Antigravity, its latest development environment that incorporates agent models and browser automation with Chrome. Building on the existing Visual Studio Code (VS Code), Antigravity provides a unified workspace for developers to work with agentic AI workflows.
To get started, users can choose from two interaction modes: Planning mode, which generates artifacts of the agent’s thinking process, or Fast mode, which executes commands directly. Antigravity comes pre-equipped with several agent models, including Gemini 3 Pro and Claude Sonnet 4.5, but paid tiers are not yet supported.
Working with the agent is relatively straightforward, as users can provide instructions to generate mockups, graphics, and even perform browser actions using a plugin or Python script. However, the outcomes may not be predictable, especially when compared to writing scripts directly.
Antigravity has its limitations, such as potential misfires, quirks specific to the IDE, and no obvious way to fix issues like Knowledge Item retention. Despite these challenges, Antigravity presents a unified workspace for developers to work on their projects, leveraging AI-driven development processes.
Key differences between Antigravity and Kiro include the models offered (Gemini vs. Claude) and the lack of developer features like hooks. While Antigravity is still in its early stages, it’s clear that Google is focusing on core agentic functions before expanding the feature set.
Source: https://www.infoworld.com/article/4096113/a-first-look-at-googles-new-antigravity-ide.html