Google releases ‘most intelligent model to date,’ Gemini 2.5 Pro
1 min read
Summary
Google has announced its most intelligent model to date, Gemini 2.5, promising enhanced reasoning and performance.
Available on Google’s AI Studio and the Gemini chat interface, with availability on Vertex AI to be introduced shortly, the model ‘thinks’ before responding to multimodal inputs, including text, audio and images, as well as coding repositories.
Google’s CTO at DeepMind, Koray Kavukcuoglu, described Gemini 2.5 as the next step in the company’s goal to develop AI that is smarter and more capable of reasoning, adding that the model represented a new level of performance by enhancing the base model and improving post-training.
The new model also enters a market that has changed a lot since the December release of Gemini 2.0, with competition from DeepSeek and OpenAI.