Last update on March 26
Today, we present Gemini 2.5, our most intelligent AI model. Our first version 2.5 is an experimental version of 2.5 pro, which is at the cutting edge of technology on a wide range of landmarks and beginnings at n ° 1 Lmarena by significant margin.
Gemini 2.5 models are models of reflection, capable of reasoning through their thoughts before responding, resulting in improved performance and improved precision.
In the field of AI, the capacity of a “reasoning” system refers to more than classification and prediction. He refers to his ability to analyze information, draw logical conclusions, integrate the context and nuances and make informed decisions.
For a long time, we have explored ways to make AI smarter and more capable of reasoning through techniques like learning to strengthen And Invitation to the chain of thoughts. By relying on this, we recently introduced our first model of reflection, Gémini 2.0 Flash Thinking.
Now, with Gemini 2.5, we have reached a new level of performance by combining a significantly improved basic model with a post-training improvement. In the future, we build these capacities of reflection directly in all our models, so that they can manage more complex problems and support even more capable and conscious agents.