Antigravity: Build Apps Faster with Autonomous Agents

Google Antigravity: The Future of AI-Powered Development Is Here

On November 18, 2025, Google launched Antigravity, a revolutionary agentic IDE powered by Gemini 3 Pro that’s redefining how developers build software. Available as a free public preview on MacOS, Windows, and Linux, Antigravity combines a VS Code fork with autonomous AI agents, multi-model support including Claude Sonnet 4.5 and GPT-OSS, and an agent-first workflow to automate planning, coding, testing, and debugging.

What Is Google Antigravity?

Google Antigravity is an agentic development platform built as a fork of Visual Studio Code, making it instantly familiar to developers. Unlike traditional coding assistants, it enables autonomous AI agents to plan, write, test, and debug entire projects with minimal human intervention, operating through direct access to the editor, terminal, and an integrated browser via a dedicated Chrome extension for end-to-end workflows and self-validation.

Key Features and Capabilities

Antigravity introduces several innovative features that set it apart from existing development tools:

• Agent-First Paradigm: Developers can delegate complex, multi-step tasks to AI agents via the Open Agent Manager, which supports asynchronous and parallel execution across multiple workspaces.
• Multi-Agent Collaboration: Multiple agents can collaborate on a single project, with one generating code while others handle testing, refactoring, or browser automation.
• Artifacts for Transparency: Agents produce tangible deliverables such as task lists, implementation plans, screenshots, browser recordings, code diffs, and walkthroughs, with all prior states preserved for review.
• Integrated Feedback: Users can leave inline feedback on artifacts like in Google Docs, adjusting review policies (e.g., Request Review, Agent Decides) so agents incorporate input seamlessly.
• Image Generation: Antigravity leverages Google’s Nano Banana (powered by Gemini 2.5) for advanced UI mockups, image editing, and generation, turning prompts into visuals and code.

Supported AI Models

Antigravity is powered primarily by Google’s Gemini 3 Pro (High and Low variants), optimized for code reasoning, large context comprehension, and multi-step planning. The platform also supports Anthropic’s Claude Sonnet 4.5 (Thinking and Standard) and open-source variants of OpenAI models like GPT-OSS 120B, giving developers flexibility.

Cross-Platform Availability

Antigravity is available for free on MacOS, Windows, and Linux, ensuring broad accessibility. The public preview includes generous rate limits for Gemini 3 Pro usage, plus options like planning mode for oversight and fast mode for quick tasks, making it ideal for individuals and teams.

How Antigravity Changes Development

Antigravity shifts the focus from manual coding to orchestrating AI agents via tools like the inbox for task management (statuses: Idle, Running, Blocked) and a knowledge base for persistent insights, allowing developers to operate at a higher, task-oriented level. By automating repetitive and complex workflows, it frees developers for creative problem-solving, with transparency via artifacts and feedback building trust.

Conclusion

Google Antigravity represents a major leap forward in AI-assisted software development. With its agent-first approach, multi-model support, and robust artifact system, Antigravity empowers developers to build software faster, smarter, and with greater confidence. As the platform continues to evolve, it promises to reshape the future of coding and open up new possibilities for innovation.