Google rolls out Gemini 2.5 Pro (I/O edition) with improved coding skills

According to Google, Gemini 2.5 Pro (I/O edition) is better than the current Gemini 2.5 Pro, especially at writing, editing, and modifying code to build interactive and visually appealing web apps

Just ahead of its I/O developer conference, Google has launched a new version of its advanced AI model — Gemini 2.5 Pro Preview (I/O edition). This update is aimed mainly at developers, but it brings improvements that could improve how users interact with apps, websites, and artificial intelligence (AI).

 

The updated AI model is available now through Google’s Gemini API, Vertex AI, AI Studio, and the Gemini chatbot app on both mobile and web. It costs the same as the previous Gemini 2.5 Pro model.

 

“Very excited to share the best coding model we’ve ever built! Today we’re launching Gemini 2.5 Pro Preview 'I/O edition' with massively improved coding capabilities. Ranks no.1 on LMArena in Coding and no.1 on the WebDev Arena Leaderboard. It’s especially good at building interactive web apps - this demo shows how it can be helpful for prototyping ideas. Try it in @GeminiApp, Vertex AI, and AI Studio http://ai.dev Enjoy the pre-I/O goodies!,” said Demis Hassabis, cofounder and chief executive officer of Google DeepMind.

 

Very excited to share the best coding model we’ve ever built! Today we’re launching Gemini 2.5 Pro Preview 'I/O edition' with massively improved coding capabilities. Ranks no.1 on LMArena in Coding and no.1 on the WebDev Arena Leaderboard.  It’s especially good at building… pic.twitter.com/9vRaP6RTTo

— Demis Hassabis (@demishassabis) May 6, 2025

What is new

 

Google says the new version is much better at tasks related to coding. This includes writing, editing, and modifying code — especially for building interactive and visually appealing web apps. It also performs better when developers need to transform code for a specific purpose.

 

According to Google, Gemini 2.5 Pro (I/O edition) even topped WebDev Arena Leaderboard, a benchmark measuring a model’s ability to create aesthetically pleasing and functional web apps. When it comes to understanding video content – a big area for AI – Google claimed that it had scored 84.8 per cent on a renowned benchmark test called VideoMME.

Google also mentioned that this update improves the model’s function calling – a key part of how AI helps developers write clean, error-free code.