Google has announced the release of Gemini 2.5, which is a new reasoning model that the company claims is its “most intelligent AI model” yet.
“Gemini 2.5 models are thinking models, capable of reasoning through their thoughts before responding, resulting in enhanced performance and improved accuracy. 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,” Koray Kavukcuoglu, CTO of Google DeepMind, wrote in a blog post.
Gemini 2.0 Flash Thinking was the company’s first reasoning model, and Gemini 2.5 builds on that with a better base model and improved post-training. In its announcement, Google revealed that all of its future AI models will have reasoning capabilities built in.
RELATED CONTENT: Mar 21, 2025: AI updates from the past week — Anthropic web search, Gemini Canvas, new OpenAI audio models, and more
The first Gemini 2.5 model is Gemini 2.5 Pro Experimental, and it leads in LMArena benchmarks significantly over other reasoning models like OpenAI o3-mini, Claude 3.5 Sonnet, and DeepSeek R1.
It also scored 18.8% on Humanity’s Last Exam, which is “a dataset designed by hundreds of subject matter experts to capture the human frontier of knowledge and reasoning.” It also excels at coding, specifically creating web apps and agentic applications, and handling code transformation. For comparison, OpenAI o3-mini scored 14% and DeepSeek R1 scored 8.6%.
This model is now available in Google AI Studio and in the Gemini app for Advanced subscribers. Google is working on adding it to Vertex AI as well, and in the next few weeks it will also announce pricing for the model.
At launch, it offers a 1 million token context window, and the company is working on adding a 2 million token context window soon.