Google has released Gemini 3.1 Pro, an upgraded core intelligence model designed for complex problem-solving, rolling it out to developers, enterprises, and consumers.
- The model achieves a verified score of 77.1% on the ARC-AGI-2 benchmark, more than doubling the reasoning performance of Gemini 3 Pro.
- It is available in preview via the Gemini API, Google Antigravity, Vertex AI, NotebookLM, and Android Studio.
- Capabilities include generating code-based animated SVGs, synthesizing complex system dashboards, and creating interactive 3D experiences.
The release aims to validate updates and advance agentic workflows before general availability.