Google is known for being the internet's largest search engine. But since Large Language Models (LLMs) gained pace, the company wants more.
As a tech giant, Google is already at the forefront of the AI revolution, competing to push the boundaries of LLMs following the rise of OpenAI's ChatGPT that started the trend by making AI disrupt industries by automating tasks, enhancing creativity, and redefining how humans interact with technology.
And as these models grow more advanced, they challenge traditional workflows, reshape economies, and force companies to adapt or risk obsolescence in the rapidly evolving AI landscape.
This time, as Google continues to innovate its AI, the company is giving it multimodal abilities and reasoning.
Just a few months after releasing Gemini 2.0 and the rise of China's DeepSeek, Google announced its "most intelligent model" yet.
Google calls it the 'Gemini 2.5,' an AI capable of reasoning and with better performance and accuracy.
Today, we’re introducing Gemini 2.5, our most intelligent AI model.
An experimental version of Gemini 2.5 Pro is available now in the Gemini app for Gemini Advanced users: https://t.co/973tlgjcKq
Let’s get into this update pic.twitter.com/AWccU7rrtb— Google Gemini App (@GeminiApp) March 25, 2025
In a blog post, Google said that:
"Gemini 2.5 models are thinking models, capable of reasoning through their thoughts before responding, resulting in enhanced performance and improved accuracy."
Gemini was the first to introduce reasoning and agentic use cases, but Gemini 2.5 takes it to an entirely new level with its exceptional reasoning and coding capabilities.
Beyond that, Gemini 2.5 Pro sets a new standard across benchmarks that demand advanced reasoning.
Think you know Gemini? Think again.
Meet Gemini 2.5: our most intelligent model The first release is Pro Experimental, which is state-of-the-art across many benchmarks - meaning it can handle complex problems and give more accurate responses.
Try it now →… pic.twitter.com/bFcx0IlY24— Google DeepMind (@GoogleDeepMind) March 25, 2025
Even without test-time techniques that drive up costs, like majority voting, 2.5 Pro excels in math and science benchmarks such as GPQA and AIME 2025.
It also achieves a state-of-the-art 18.8% on Humanity’s Last Exam—an expert-designed dataset that tests the boundaries of human knowledge and reasoning—without relying on external tools.
In all, this Gemini 2.5 is Google’s goal of making "AI smarter and more capable of reasoning."
As a thinking model, Gemini 2.5 has enhanced performance, improved accuracy, and the ability to support even more capable, context-aware agents.
This means it’s well-suited for:
Multimodal Reasoning
Advanced Coding
Math & Science pic.twitter.com/OWCQiFQIz9— Google Gemini App (@GeminiApp) March 25, 2025
"Now, with Gemini 2.5, we’ve achieved a new level of performance by combining a significantly enhanced base model with improved post-training,” Google said.
"Going forward, we’re building these thinking capabilities directly into all of our models, so they can handle more complex problems and support even more capable, context-aware agents."
Google's 2.0 Pro model was already impressive when it first launched. But now with Gemini 2.5 being released, the older Gemini 2,0 Pro if no more.
Google says that Gemini 2.5 Pro is a drop-in replacement for 2.0 that shall be available across Google's products for anyone with a Gemini Advanced subscription.
Initially, Google is making this model available in Gemini mobile app and on the web, as well as in Google's AI Studio.
Google One AI Premium subscription required. Gemini Advanced and Gemini for Gmail, Docs, and more are only available for ages 18+. Certain Gemini Advanced features and Gemini for Gmail, Docs and more are only available in select languages.
— Google Gemini App (@GeminiApp) March 25, 2025