Last week, Google made a significant update to Gemini 3 Deep Think and officially launched the upgraded core model, Gemini 3.1 Pro, today. This model has been fully integrated into Google's consumer and developer products. Starting today, Gemini 3.1 Pro will be gradually made available to developers, enterprise users, and general users. Based on the Gemini 3 series, the core reasoning capabilities of 3.1 Pro have been significantly enhanced, achieving a score of 77.1% on the ARC‑AGI‑2 benchmark test, with reasoning performance more than double that of the previous generation 3 Pro. Designed specifically for handling complex tasks, this model can be applied to various real-world scenarios. Google AI Pro and Ultra subscribers enjoy higher usage limits when using 3.1 Pro in the Gemini app, while NotebookLM exclusively offers 3.1 Pro features to Pro and Ultra users. Developers and enterprise users can experience its preview version.