Google has launched an early preview of Gemini 2.5 Pro (I/O edition), offering major improvements for developers building interactive web apps. Originally planned for release at Google I/O, the update is now available early due to strong demand.
This new version greatly enhances coding performance, especially in web development tasks. It also improves at transforming code, editing existing scripts, and supporting advanced workflows.
Gemini 2.5 Pro now leads the WebDev Arena Leaderboard, outperforming the older version by 147 Elo points based on user preferences for app design and functionality. It also excels in multimodal tasks, with a top score of 84.8% in video understanding on the VideoMME benchmark.
Developers can use the updated model through Google AI Studio, Vertex AI, and the Gemini app, which includes tools like Canvas for visual coding. Google says the model is now better at calling tools during coding, reducing failure rates in real-world use.
With stronger performance across UI building, long context handling, and video input, Gemini 2.5 Pro Preview is now one of the most capable AI models for interactive app development.
