Google has introduced Gemini 2.5 Pro, its latest innovation in the field of artificial intelligence. This model is considered the most powerful AI model in the company's history and demonstrates significant advancements compared to previous versions. Gemini 2.5 Pro excels in data analysis, programming, and solving complex problems. It can process up to 1 million tokens of context, with the capacity to expand to 2 million tokens.
This new model offers substantial improvements over its predecessors. Notably, its enhanced reasoning capabilities stand out; the model can construct a logical chain before providing a final answer. Users also have the option to manually activate the "thinking out loud" mode.
Gemini 2.5 Pro has achieved first place in the LMArena ranking and set record-breaking results in tests such as AIME 2025 (mathematics), GPQA diamond (science), and Humanity’s Last Exam. In the SWE-Bench Verified programming test, the model scored 63.8%, reflecting its exceptional performance in coding tasks.
The model features multimodal capabilities, allowing it to work with various data types, including text, images, videos, audio files, and code repositories. Soon, Gemini 2.5 Pro will be available on the Vertex AI platform, and Google plans to announce its pricing policy for corporate use. Currently, the model is accessible in trial mode for subscribers and developers.
Google's CEO, Sundar Pichai, has described Gemini 2.5 Pro Experimental as the "most advanced reasoning model", stating that it will set new standards in the AI industry.
