Google Launches Gemini 3.1 Pro With Major Reasoning Upgrade Ahead of Apple's Siri Overhaul

Google has announced Gemini 3.1 Pro, the latest version of its flagship AI model. The new release is built for demanding tasks, with a focus on complex logic, coding, and multi-step problem-solving.



The update also carries implications for the Apple ecosystem. Apple recently solidified a multi-year partnership to utilize Gemini models to power upcoming AI features. While Apple handles basic requests locally on the iPhone and Mac, advanced models like Gemini 3.1 Pro are expected to serve as the cloud-based foundation for a major Siri overhaul slated for iOS 27, transforming the assistant into a more conversational system capable of executing complex workflows.


According to Google, Gemini 3.1 Pro is designed for tasks where straightforward answers are not enough. The company says the model can synthesize large volumes of data into a unified view and generate clear explanations of complex topics. In internal testing, Gemini 3.1 Pro achieved a verified score of 77.1 percent on the ARC-AGI-2 benchmark, which evaluates an AI's ability to solve novel logic patterns. Google says that score is more than double the reasoning performance of Gemini 3 Pro.

Benchmark data released alongside the announcement shows the model leading several top-tier offerings from industry rivals. In academic reasoning and scientific knowledge tests such as Humanity's Last Exam and GPQA Diamond, Gemini 3.1 Pro posts leading results, outperforming Anthropic's Claude 4.6 Opus and OpenAI's recently launched GPT-5.2. It also performs competitively in agentic coding evaluations like SWE-Bench Verified, as companies continue pushing toward more automated software development.

Gemini 3.1 Pro is rolling out in preview today across Google's consumer and developer platforms. For users, it is available in the Gemini app with higher usage limits for subscribers on the Google AI Pro and Ultra plans. It is also accessible in NotebookLM exclusively for Pro and Ultra users.

Developers can test the model in preview through the Gemini API in Google AI Studio, the Gemini CLI, Android Studio, and the agentic development platform Google Antigravity. Enterprise customers can access it through Vertex AI and Gemini Enterprise as Google works toward general availability.