New model doubles prior ARC-AGI-2 score and rolls out across Gemini app, NotebookLM, Vertex AI and developer tools.
Google has released Gemini 3.1 Pro, an upgraded version of its flagship model designed to significantly improve core reasoning and complex problem solving across consumer, enterprise and developer products.
The company said Gemini 3.1 Pro more than doubles the ARC-AGI-2 benchmark score of Gemini 3 Pro, achieving a verified 77.1% on the test that measures a model’s ability to solve entirely new logic patterns. The improvement positions the model for workflows that require multi-step reasoning rather than simple responses.
Gemini 3.1 Pro is rolling out in preview for developers through the Gemini API in Google AI Studio, Gemini CLI, Google Antigravity and Android Studio. Enterprise users can access it via Vertex AI and Gemini Enterprise, while consumers will see it in the Gemini app and in NotebookLM for Google AI Pro and Ultra subscribers.
Google said the upgrade builds on last week’s Gemini 3 Deep Think update, which targeted scientific, research and engineering use cases. The company plans to use the preview phase to refine ambitious agentic workflows before broader general availability.
