"Google's smartest AI model yet": the Gemini 2.5 Pro with deep thinking capability unveiled
Google has unveiled Gemini 2.5 Pro, an advanced AI model with deep "thinking" features.
Here's What We Know
Gemini 2.5 Pro is the analogue of OpenAI's o1 model and, according to Google, will be the basis for the development of all future Gemini models.
"Reasoning" (Reasoning) allows Gemini 2.5 Pro to deeply analyse information, draw logical conclusions and solve complex problems, taking context into account.
Google believes that in the future, such AI will be able to perform tasks autonomously. The context window of Gemini 2.5 Pro is 1 million tokens, and they plan to double it in the future.
In tests, Gemini 2.5 Pro shows significant improvements over past generations and ranked first (18.8%) in the Humanity's Last Exam test, which evaluates AI knowledge and reasoning at the human limit.
Humanity's Last Exam test results
The Gemini 2.5 Pro Experimental Preview is available in AI Studio and for Gemini Advanced subscribers. Cost has not yet been disclosed.
Source: Google Blog
