Gemini 2.5 Pro Revealed: Google’s Next-Gen AI Takes Center Stage Before I/O 2025

This month, Google is getting ready to host Google I/O 2025, its annual developer conference. However, the tech giant has unveiled a new iteration of its AI model, Gemini, ahead of its annual software presentation.

 The company's flagship Gemini 2.5 Pro AI model has been enhanced and is now available as the Gemini 2.5 Pro Preview.

With the intention of giving developers early access to its improved capabilities, the business unveiled the new model—named Gemini 2.5 Pro Preview (I/O edition)—earlier than anticipated. Google claims that the Gemini 2.5 Pro Preview has enhancements in a variety of code areas.

Google Claims: Best Coding Model Ever Made

Demis Hassabis, the CEO of Google DeepMind, wrote on X about the new AI model's release. It is the best coding paradigm ever created, according to Hassabis. He goes on to say that the concept is perfect for developing interactive applications.

Additionally, Haasabis posted a video demonstrating how to use Gemini 2.5 Pro Preview for idea development. Haasabis expressed his excitement about sharing the best coding model brand ever created in a post on X.

The company is releasing the Gemini 2.5 Pro Preview 'I/O edition' today, which has significantly enhanced coding capabilities. Holds the top spot on the WebDev Arena Leaderboard and the LMArena Coding Ranking.

 It excels at creating interactive web applications; this example demonstrates how useful it is for concept prototyping. Try it in AI Studio (http://ai.dev), Vertex AI, and @GeminiApp. Savour the pre-I/O treats.

New Improved Features of Gemini 2.5 Pro

Google claims that Gemini 2.5 Pro Preview offers significant advancements in several coding areas, including front-end and user interface development, code editing and transformation, and the building of complex agentic workflows.

It is anticipated that these improvements will expedite the development process, allowing programmers to create more inventive and effective apps.

According to a blog post by Google, developers who are now using Gemini 2.5 Pro will benefit from improved coding performance as well as crucial developer feedback, such as fewer function calling errors and higher function calling trigger rates.

No action is necessary to use the enhanced model, which is still available at the same price, as the prior iteration (03-25) now redirects to the more recent version (05-06).

Gemini 2.5 Pro achieves 84.8% on the VideoMME standard, demonstrating cutting-edge video comprehension. When combined with code, this allows for new flows that were previously unattainable with earlier iterations.

The Video to Learning App in Google AI Studio, for instance, shows how Gemini 2.5 Pro uses a single YouTube video to generate an interactive learning application.

Compared to the prior basic example, the new Gemini 2.5 Pro model offers a more useful experience with enhanced video understanding and a comprehensive user interface.