Gemini 3.1 Pro, Building with Advanced Intelligence in Google Antigravity

Since we first introduced Google Antigravity, our mission has been to move beyond simple code completion and toward an era of agent-first development. Today, we’re taking a step forward by bringing Gemini 3.1 Pro directly into your Antigravity workflow.

Expanding on the Gemini 3 series, 3.1 Pro marks a significant advancement in fundamental reasoning. As a smarter and more adept baseline for intricate problem-solving, 3.1 Pro is built for scenarios where basic responses fall short. It's strong in complex tasks, like workflows that require the model to think through the optimal path for a solution.

Whether you are refactoring a legacy authentication module or building immersive and interactive data visualizations, Gemini 3.1 Pro works intelligently and efficiently, thinking more deeply to solve an even wider range of coding problems.

In Antigravity, this translated intelligence means the agents are more capable of:

• Robust planning: Mapping out a complete architectural plan before touching a single line of code.
• Code-based animation: Generate website-ready animated SVGs from a text prompt built to scale in pure code rather than pixels.
• Navigating long-horizon tasks across entire codebases while maintaining perfect context.
• Dynamic visualization: Turning high-level concepts into interactive UI components and dynamic data visualizations that improve your user experience.

The upgraded reasoning in 3.1 Pro translates directly to higher productivity for developers. To showcase this, we asked Gemini 3.1 Pro to navigate an intricate database migration. It generated a comprehensive risk assessment where it proactively addresses potential data loss by architecting a Local-First sync engine and a robust conflict resolution strategy. This ensures the application remains resilient and user data stays protected throughout the transition:

We can also see how Gemini 3.1 Pro uses advanced reasoning to transform dense research into functional code. By ingesting complex PDFs directly into Antigravity, the agent translates high-level theory into an interactive Local-First simulation. It successfully architects a CRDT-backed network, complete with real-time sync and connection toggling, bridging the gap between academic papers full of high-level theory and a debugged, interactive web application in a single long-horizon task:

Gemini 3.1 Pro is now available in the latest version of Google Antigravity.

In addition to Google Antigravity users, Gemini 3.1 Pro is rolling out globally starting today via the Gemini API in Google AI Studio, Vertex AI, Gemini CLI, and Android Studio for developers.

Experience these new capabilities by selecting Gemini 3.1 Pro in Antigravity: