NEW YORK, April 2 — Google has now launched Gemini 2.5 Pro, its most advanced artificial intelligence model to rival ChatGPT 4.5. This model has been designed to analyse complex information, draw logical conclusions and provide answers that are both more accurate and more nuanced than before.

Compared to previous versions, Gemini 2.5 Pro boasts much improved performance.

The most spectacular advances concern programming.

The model can now generate and correct code from extremely simple requests.

For example, Gemini can now develop an entire application or even a video game based on simple prompts and a preferred language.

In addition, its ability to simultaneously process several types of file (text, images, audio, video, code, etc.), combined with a context window soon extended to two million tokens, will facilitate the analysis of voluminous data and the resolution of complex problems, including those of a mathematical nature.

Note that a token represents an elementary unit of data. In the context of language models, this could be a part of a word, for example.

In all cases, Gemini systematically displays its current reasoning, enabling users to refine their prompts over time.

Today, all Gemini users have access to Gemini 2.5 Pro, although usage is limited in the free version. A mobile version is due to be deployed shortly.

This launch comes on the heels of a huge buzz generated by OpenAI with the launch of a version of its 4o model dedicated to image creation. This has since been used by a huge number of internet users to generate pictures inspired by artists such as Hayao Miyazaki or Matt Groening. — ETX Studio