The new version of Google’s Artificial Intelligence was released with the improvement of the programming ability

Google has unveiled a advanced and advanced version of its flagship artificial intelligence, Gemini 2.5 Pro last night. Known as the unofficial Gemini 2.5 Pro (I/O Edition, this version has been made public before the annual Google Developers I/O conference, bringing significant improvements, especially in the field of coding. The move reflects Google’s continuous efforts to maintain its leading position in the intense competition of the artificial intelligence world, where competitors such as Openai and XAI are also developing and presenting their powerful models.

One of the most prominent features of the Gemini 2.5 Pro (I/O Edition) is its upgraded capabilities in the field of coding and developing interactive web applications. Google emphasizes that this new model performs much better in tasks such as converting code (changing a code section to achieve a specific goal) and editing the code. These advances are a great news for developers looking for more efficient tools for their projects.

To prove these claims, Google cites the results of valid benchmarks. In the Webdev Arena LeaderBoard ranking, which measures the ability of a model to create beautiful and efficient web applications, the Gemini 2.5 Pro (I/O Edition) has won 147 more points than its previous version. This achievement reflects Google’s focus on developing models that are not only technically powerful, but can also provide high -quality outputs and desirable user experience.

In addition, in the LiveCodebench V5 benchmark, which specifically focused on code production, the model ranging from 70.4 % to 75.6 percent. This increase in rating is a testament to the significant improvement in the accuracy and efficiency of the model in the production of the correct and optimized pieces. In the field of video content understanding, Gemini 2.5 Pro (I/O Edition) has a performance at the State-OF-THE-AT-Rate at 84.8 %.

The remarkable thing for current Gemini 2.5 Pro users is that to take advantage of these new features, there is no need to take any extra action or pay more. Google has announced that users of the old version are automatically referred to the updated model at the same cost. The update, released about two weeks earlier than the forecast at the I/O conference, has been available through the Jammeh app as well as the Google AI Studio, Vertex AI and Gemini API platforms. Developers can also use this new model using Canvas.

Google points out in its blog: “For developers who used to use Gemini 2.5 Pro, this new version not only improves coding performance, but also respond to key developers’ feedback, including reducing errors in calling functions and improving functions activation rates.” “By default, this model has a real desire to develop beautiful and attractive web, while maintaining its guidance capability,” the company also adds. These statements reflect Google’s commitment to listen to the needs of the developer community and provide tools that are both powerful and user -friendly. Early release of this version is a strategic step by Google to enhance its artificial intelligence ecosystem and attract more developers to its platforms.

Source: 9TO5MAC