Google Launches First Gemini 2.5 Model

Google today announced the initial model in its Gemini 2.5 family of AI models, describing it as its most intelligent yet.

“Gemini 2.5 is a thinking model, designed to tackle increasingly complex problems,” Google Deepmind CEO Koray Kavukcuoglu explains. “Our first 2.5 model, Gemini 2.5 Pro Experimental, leads common benchmarks by meaningful margins and showcases strong reasoning and code capabilities.”

This isn’t Google’s first reasoning model–that would be Gemini 2.0 Flash Thinking Experimental–but Gemini 2.5 Pro Experimental points the way to the future, as all Gemini 2.5 models will include reasoning capabilities.

“For a long time, we’ve explored ways of making AI smarter and more capable of reasoning through techniques like reinforcement learning and chain-of-thought prompting,” Kavukcuoglu continues. “‘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 … resulting in enhanced performance and improved accuracy.”

In addition to its reasoning capabilities, Gemini 2.5 Pro Experimental offers native multimodality and a long 1 million token context window, with 2 million coming soon. And Google says it offers strong performance improvements over previous Gemini generations. “It can comprehend vast datasets and handle complex problems from different information sources, including text, audio, images, video and even entire code repositories,” Kavukcuoglu says.

Where the Gemini 2.0 models are broadly available, Gemini 2.5 Pro Experimental is available only to those who subscribe to Gemini Advanced, its $20-per-month subscription offering, and via Google AI Studio for developers.