Gemini 2.5: Google unveils “most intelligent AI model” to rival DeepSeek, ChatGPT

Tech giant Google has released its latest Artificial Intelligence model, Gemini 2.5, a thinking version designed to tackle increasingly complex problems. The announcement is coming amid a tense competition with other giants, especially the Chinese option, DeepSeek and Sam Altman’s OpenAI.

This first 2.5 release is an experimental version of 2.5 Pro, and it’s designed to tackle increasingly complex problems. This is a significant step forward as Gemini 2.5 is Google’s most intelligent AI model to date.

“With Gemini 2.5, we’ve achieved a new level of performance by combining a significantly enhanced base model with improved post-training. 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”, Koray Kavukcuoglu, CTO of Google DeepMind, said in a blog post.

In the field of AI, a system’s capacity for “reasoning” refers to more than just classification and prediction. It refers to its ability to analyze information, draw logical conclusions, incorporate context and nuance, and make informed decisions.

According to Google, Gemini 2.5 models can reason through their thoughts before responding, resulting in enhanced performance and improved accuracy. It builds on what makes Gemini models great — native multimodality and a long context window.

The 2.5 Pro ships today with a 1 million token context window (2 million coming soon), with strong performance that improves over previous generations. It can comprehend vast datasets and handle complex problems from different information sources, including text, audio, images, video and even entire code repositories.

How Gemini 2.5 Pro compares

Gemini 2.5 Pro Experimental is Google’s advanced model for complex tasks. It tops the LMArena leaderboard by a significant margin, indicating a highly capable model equipped with high-quality style. The board measures human preferences of AI models.

It also shows strong reasoning and code capabilities, leading in common coding, math and science benchmarks. Below is a table indicating the model’s Multi-Round Coreference Resolution evaluations:

Enhanced reasoning: Without test-time techniques that increase cost, like majority voting, 2.5 Pro leads in math and science benchmarks like GPQA and AIME 2025. It also scores an 18.8% across models without tool use on Humanity’s Last Exam, a dataset designed by hundreds of subject matter experts to capture the human frontier of knowledge and reasoning.

Advanced coding: Gemini 2.5 can create visually compelling web apps and agile code applications, along with code transformation and editing. On SWE-Bench Verified, the industry standard for agentic code evals, Gemini 2.5 Pro scores 63.8% with a custom agent setup.

Increasing capacity to compete

Google is constantly updating Gemini’s features to ensure that it competes favourably in an increasingly competitive AI space. Recall that last week, it introduced Canvas and Audio Overview on its Gemini AI suite to enhance AI-driven content creation.

Canvas is a new interactive workspace within the app that simplifies document and code creation. Users can now write, edit, and refine their work in real time, with Gemini providing intelligent feedback and editing suggestions.

Also with Canvas, users can generate and preview HTML/React code, modify input fields, and add buttons with instant updates. This functionality streamlines the coding process, reducing the need to switch between multiple applications.

Similarly, Audio Overview converts documents, slides, and research reports into dynamic, podcast-style audio discussions.

“Gemini will create a podcast-style discussion between two AI hosts who, with just a click, launch into a lively deep-dive conversation based on your uploaded files,” Google explained. These AI hosts summarize content, connect key topics, and provide unique insights.

Gemini 2.5 Pro is available in Google AI Studio and the app for Advanced users and will be coming to Vertex AI. Developers and enterprises can start experimenting with Gemini 2.5 Pro in Google AI Studio, and Advanced users can select the model dropdown on desktop and mobile.