Google announces Gemini 2.5 AI capable of reasoning before answering

Google has announced Gemini 2.5: Our newest Gemini model with thinking, the company’s most intelligent AI.  Gemini 2.5 is launching in an experimental mode for developers in the AI Studio, and it’s also available for Gemini Advanced users in the Gemini app. You can access this mode in the dropdown menu on both desktop and mobile.

Gemini AI can create “visually compelling web apps and agentic code applications.” Google shows off this little create-your-own dinosaur game and how easy it was to do with Gemini 2.5 below.

Google is launching Gemini 2.5 Pro first. This AI is capable of reasoning through thoughts before answering—which is better than a lot of humans I know. Gemini 2.5 will have a one million token context window at launch, expanding to two million.

The Mountain View company says that they created Gemini 2.5 by combining an enhanced base model and then improved it with post-training. This allows it to outperform other popular AI models and gives it the ability of reasoning, science and mathematics.

Source: Google Blog