Google launches Gemini 2.5 Pro Preview to boost AI coding and video analysis

Google has launched the Gemini 2.5 Pro Preview (I/O edition), an updated version of its flagship AI model. This new model is available via Gemini API, Vertex AI, and AI Studio platforms. It also integrates with the Gemini chatbot app on both web and mobile devices.

The release precedes Google’s annual I/O developer conference.

Advancements in AI Coding and Web Development

The Gemini 2.5 Pro Preview focuses on improved coding and web application development. It tops the WebDev Arena Leaderboard, which rates AI models on creating functional and attractive web apps. This shows the model’s advanced ability in building interactive web experiences.

By addressing developer feedback, Google reduced errors in function calling and improved trigger rates. The model defaults to prioritizing aesthetic web design while remaining versatile for various tasks.

The model scored 84.8% on the VideoMME benchmark, demonstrating stronger video understanding capabilities. Google’s enhancements reflect its strategy to empower developers with better AI tools.

• Enhanced Coding: Excels at code transformation, editing, and interactive app building.
• Video Understanding: Achieved 84.8% on VideoMME, indicating improved video content analysis.
• Extended Context: Processes up to 1 million tokens for comprehensive data analysis.

This updated AI is now free to all users, as highlighted in a recent announcement by Tom’s Guide. This move expands accessibility for developers and researchers.

Google’s Strategy in AI Competition

The release of Gemini 2.5 Pro Preview is part of Google’s effort to compete with companies like OpenAI and xAI. The updated model offers stronger performance and features to gain market share.

Developers can access the model through multiple platforms, including Google Cloud at events like Google Cloud Next 2025. These platforms ensure broad usability of the AI.

The improvements target key developer concerns, enhancing reliability and function calling accuracy. Integration with the Gemini chatbot app improves user interaction across devices.

Furthermore, the model supports expansive projects by handling large contexts, enabling developers to analyze extensive datasets efficiently.