Google has just released early access to Gemini 2.5 Pro Preview (I/O edition) an enhanced version of its AI model that brings major improvements for developers, especially those building interactive web apps.

Although it was originally planned for release during Google I/O in a few weeks, Google decided to launch it early because of the huge interest and excitement from the developer community.

What’s New in This Update?

The new version builds on the strong performance of Gemini 2.5 Pro, with major upgrades in,

Web development: Now better at creating visually appealing and functional web apps.

Code editing and transformation: Improved ability to edit, refactor, and rewrite code.

Advanced workflows: Can handle more complex agent-based tasks.

These enhancements have helped the model climb to the top of the WebDev Arena Leaderboard, where it leads with a +147 Elo point gain over the previous version. This leaderboard is based on how much people prefer one model over another in building web applications.

See the result of this prompt in the Gemini app.

Strong in Multimodality and Long Context

Gemini 2.5 Pro Preview also continues to lead in other areas.

Multimodal understanding: Processes images, videos, and text together naturally.

Video analysis: Achieved 84.8% on the VideoMME benchmark, showing its strong video understanding abilities.

Long-context processing: Handles larger inputs and conversations more effectively.

Try It Now

The new Gemini 2.5 Pro Preview is available to try in the Gemini app, so developers can start exploring its new features right away.