Unlocking the Power of Gemini 2.5 Pro: A Developer's Guide

Published On Thu May 08 2025

Home - News - Google releases Gemini 2.5 Pro — AI model creates web applications from simple ideas

Google releases Gemini 2.5 Pro — AI model creates web applications from simple ideas

Google continues to develop and improve its Gemini AI assistant. After the release of the model Gemini 2.0 Flash optimized for daily activities and other tasks, the company announced the launch of Gemini 2.5 Pro Preview for developers. Interestingly, it was announced ahead of schedule.

Initially, the new Gemini 2.5 Pro was planned to be presented during the Google I/O conference on May 20-21. However, due to the extremely high interest of the developer community, the company decided to unveil the model earlier.

Google said, "Today we’re providing early access to Gemini 2.5 Pro Preview (I/O edition), an updated version of 2.5 Pro that significantly improves coding capabilities, especially for creating immersive, interactive web apps."

Improved Coding Capabilities

The new version of Gemini 2.5 Pro Preview was developed based on the feedback from the first release Gemini 2.5 Pro. In the WebDev Arena ranking, the updated model outperformed the previous one by 147 points on the Elo scale.

Google notes that Gemini 2.5 Pro understands code well and is able to reason logically, making it suitable for developers. The model also possesses strong video understanding capabilities, scoring 84.8% on the VideoMME test.

Practical Use and Partnerships

Gemini 2.5 Pro is already integrated into innovative tools such as the Cursor code agent and is used in partnerships with companies like Cognition and Replit. Silas Alberti from Cognition stated, "It was like working with a more experienced programmer — the model made the right assumptions and chose good abstractions." Replit’s Michele Catasta added that they consider Gemini 2.5 Pro to be the best advanced model in terms of the feature-to-latency ratio.

In its blog post, Google provided examples of the practical use of the model. For instance, it can turn YouTube videos into full-fledged educational applications, generate stylish UI components, and create web applications from simple ideas.

Availability and Usage

Developers can already access the new version of Gemini 2.5 Pro via Google AI Studio or Vertex AI. If you have used the previous version, you will be automatically redirected to the latest version without the need for additional steps or payments.

Previously, Google added to the Gemini app its experimental artificial intelligence model that can "reason" and explain how it answers complex questions.