Google has a metal pedal in the development of its AI. Just a few months after the debut of GEMINI 2.0The tech giant opened another upgrade to Gemini 2.5. Like any new AI launch, Google has posted a strong Lmarena performance for Gemini 2.5, especially its capabilities in coding, math and science.
The first model in this series is the Gemini 2.5 pro experimental. Google said this is a model of thought intended to provide responses based on more reasoning, review and context than the answers offered by the classification- and the prophecy-driven-driven models. This is a different approach than Google took along with the Gemini 2.0 series, which started with a better and less expensive version of Flash.
"In Gemini 2.5, we achieved a new level of performance by incorporating a significantly enhanced base model with enhanced post-training," The company said in a blog post linked to Koray Kavukcuoglu, CTO of Google Deepmind. "Continuing, we are building these thinking capabilities directly in all of our models, so that they can handle more complex problems and support more capable, contextual awareness agents."
Google just started the Gemini 2.0 rotation in its services, using it to empower the newly added AI mode in search and Deep research For handling more complex queries. In today's launch, expect to hear more updates from the company about getting the latest version. The Gemini 2.5 Pro Experimental is now available at Google AI Studio, and advanced Gemini members can use it directly with the Gemini app.
This article originally appeared on Engadget at https://www.engadget.com/ai/google-releases-gemini -25-ai-modeel-for-complex-ladking-182352224.html? SRC = RSS
