Google has introduced a new family of ‘reasoning’ AI models, Gemini 2.5, which pause before responding to allow for contemplation.
Introducing Gemini 2.5, our most intelligent AI model.
Our first release, an experimental version of 2.5 Pro, unlocks state-of-the-art performance in math and science. ?
Learn more ? pic.twitter.com/aoe7egliJb
— Google (@Google) March 25, 2025
The first in the series is the experimental version of Gemini 2.5 Pro. This multimodal neural network with reasoning capabilities is described as ‘the most intelligent to date.’ It is available on the Google AI Studio platform and in the Gemini app for Advanced subscription holders at $20 per month.
“In Gemini 2.5, we have reached a new level of performance by combining a significantly improved base model with enhanced post-training,” the corporation’s blog states.
The company added that in the future, all its models will possess reasoning capabilities.
In the Aider Polyglot test, which tracks programming skills, Gemini 2.5 Pro scored 68.6%, surpassing leading solutions from OpenAI, Anthropic, and DeepSeek.
In the SWE-bench Verified benchmark, which measures software developers’ capabilities, Google’s solution was outperformed by Claude 3.7 Sonnet—63.8% versus 70.3%.
In the multimodal test Humanity’s Last Exam, consisting of a thousand questions in mathematics, humanities, and natural sciences, Gemini 2.5 Pro scored 18.8%, exceeding competitors’ results.
Gemini 2.5 Pro has a context window of 1 million tokens—the model can perceive approximately 750,000 words at once. There are plans to double this figure in the future.
‘Reasoning’ AI requires more time and computational resources to provide accurate and comprehensive responses. The first ‘thinking’ neural network was introduced by OpenAI—in September 2024, it released o1. Other AI developers have since embarked on creating their own solutions, including DeepSeek, Anthropic, Alibaba, and others.
In March, Google announced a range of AI products and features for the healthcare sector.
Previously, the company trained its AI assistant Gemini to analyze and consider a user’s search history to provide more relevant responses.