Google announces its newest and most intelligent Gemini AI model.

Gemini 2.5 is considered Google's newest and most intelligent AI (artificial intelligence) model to date. The first version of the 2.5 generation is Gemini 2.5 Pro – a model capable of reasoning through pre-thinking before responding, similar to Gemini 2.0 Flash Thinking. Gemini 2.5 Pro will come with a contextual window of up to 1 million tokens, and Google says this number will soon be expanded to 2 million.

Comparing the power of Gemini 2.5 Pro with some current AI models.

According to Google, Gemini 2.5 significantly outperforms many current popular AI models in areas such as reasoning, science , and mathematics. The company achieved this by "combining a significantly improved base model with optimized post-training."

Some key features of Gemini 2.5

One of the highlights of Gemini 2.5 is its enhanced coding capabilities, enabling the creation of visually appealing web applications as well as command-based applications. To demonstrate this capability, Google shared a video showing the AI ​​creating a game based on text prompts.

What would a smart city look like with AI technology?

Gemini 2.5 is currently being launched as a test model, available to developers in AI Studio and Gemini Advanced users in the Gemini app. If you are a Gemini Advanced member, you can select Gemini 2.5 from the model dropdown list on both desktop and mobile apps. Google also stated that Gemini 2.5 will be available on Vertex AI in the coming weeks, however, specific details have not yet been released by the company.

Source: https://thanhnien.vn/google-cong-bo-mo-hinh-ai-gemini-moi-va-thong-minh-nhat-185250326070115295.htm