Scaling core intelligence from the research lab to your everyday workflow.
- A Massive Reasoning Jump: Gemini 3.1 Pro more than doubles the reasoning performance of its predecessor, scoring a verified 77.1% on the rigorous ARC-AGI-2 logic benchmark.
- Universal Availability: The model is rolling out today across the entire ecosystem, including the Gemini app, NotebookLM, Google AI Studio, and the “Antigravity” agentic development platform.
- Built for Complexity: Designed to bridge the gap between simple queries and deep technical breakthroughs, 3.1 Pro excels at synthesizing data, creative execution, and complex problem-solving.
The landscape of artificial intelligence is shifting from models that simply “know” things to models that can “think” through them. Following last week’s major update to Gemini 3 Deep Think—which targeted specialized breakthroughs in science and engineering—Google has now unveiled the engine behind that progress: Gemini 3.1 Pro. This update represents a significant leap in core intelligence, transforming the baseline experience for developers, enterprises, and everyday consumers alike.
At the heart of Gemini 3.1 Pro is a focus on complex reasoning. While the previous version laid a strong foundation, the 3.1 Pro model is built for the moments when a simple answer isn’t enough. This isn’t just a marginal improvement; the data shows a staggering evolution in logic. On the ARC-AGI-2 benchmark—a test specifically designed to see if an AI can solve entirely new logic patterns it hasn’t seen during training—3.1 Pro achieved a score of 77.1%. To put that in perspective, that is more than double the reasoning performance of the standard Gemini 3 Pro.
For the modern professional or creator, this means “Intelligence Applied.” Whether you are trying to synthesize a mountain of disparate data into a single, coherent view or you need a visual explanation of a dense, multi-layered topic, 3.1 Pro is designed to handle the heavy lifting. It moves beyond the role of a chatbot and into the role of a collaborator, capable of bringing ambitious creative projects to life with a more nuanced understanding of the user’s ultimate goal.
The rollout of this model is intentionally broad, ensuring that this new level of intelligence is accessible where work actually happens. Developers can begin experimenting with 3.1 Pro in preview via the Gemini API, the Gemini CLI, and Android Studio. Furthermore, it is integrated into Google Antigravity, the new agentic development platform designed for building autonomous AI workflows. Enterprise users will see the model integrated into Vertex AI and Gemini Enterprise, while consumers can access it via the Gemini app and NotebookLM.
As we look toward the future, the “preview” status of this release is a deliberate move to gather feedback on high-stakes tasks and agentic workflows before a general release. For those on the Google AI Pro and Ultra plans, the wait is already over—higher limits and exclusive access in NotebookLM are rolling out now. Gemini 3.1 Pro isn’t just a faster version of what came before; it’s a smarter, more resilient partner for the world’s most complex tasks.
