Google has rolled out early access to Gemini 2.5 Pro Preview (I/O edition), an enhanced version of its advanced AI model, ahead of the official Google I/O 2025 launch. The update is designed to elevate coding workflows, particularly in building interactive web applications.
Tulsee Doshi, Senior Director of Product Management at Google, revealed that the preview was originally scheduled for release during Google I/O. However, due to strong developer interest and growing momentum around Gemini, the company opted for an early release to give developers a head start.
The updated model improves upon the original Gemini 2.5 Pro’s performance in key areas of software development:
- Code transformation and editing
- UI design and development
- Complex agentic workflow building
- Multimodal understanding and long-context reasoning
Gemini 2.5 Pro Preview now leads the WebDev Arena Leaderboard, beating the previous model by 147 Elo points, a significant leap based on human evaluations of web app aesthetics and functionality. It also maintains strong multimodal capabilities, scoring 84.8% on the VideoMME benchmark, a test for video comprehension tasks.
Developers can access Gemini 2.5 Pro Preview through the Gemini API on Google AI Studio and Vertex AI. The model is also integrated into the Gemini app, supporting tools like Canvas, which lets users generate code and design interactive web apps using just a prompt.