Google on Tuesday announced the launch of Gemini 2.5 Pro Preview (I/O edition), an updated version of its flagship Gemini 2.5 Pro AI model that the company claims tops a number of widely used benchmarks.
Gemini 2.5 Pro Preview (I/O edition) is available via the Gemini API and Google’s Vertex AI and AI Studio platforms, and is priced the same as the Gemini 2.5 Pro model it effectively replaces. It’s also in Google’s Gemini chatbot app for the web and for mobile devices.
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
The model’s release comes ahead of Google’s annual I/O developer conference (hence the “I/O edition” designation), where Google is expected to unveil a host of models, as well as AI-powered tools and platforms. The company is competing fiercely for mindshare and marketshare in the cutthroat AI race; rivals like OpenAI and xAI are on the cusp of releasing models that are expected to be highly performant.
According to Google, Gemini 2.5 Pro Preview (I/O edition) has “significantly” improved capabilities for coding and building interactive web apps. The model is also better at tasks like code transformation — that is, modifying a piece of code to achieve a specific goal — and code editing, the company says.
In a blog post, Google notes that Gemini 2.5 Pro Preview (I/O edition) leads the WebDev Arena Leaderboard, a benchmark measuring a model’s ability to create aesthetically pleasing and functional web apps. Gemini 2.5 Pro Preview (I/O edition) also has state-of-the-art performance in video understanding, achieving a score of 84.8% on one popular benchmark, VideoMME.
“For developers already using Gemini 2.5 Pro, this new version will not only improve coding performance but will also address key developer feedback including reducing errors in function calling and improving function calling trigger rates,” wrote Google in its blog post. “By default, the model has a real taste for aesthetic web development while maintaining its steerability.”
Kyle Wiggers is TechCrunch’s AI Editor. His writing has appeared in VentureBeat and Digital Trends, as well as a range of gadget blogs including Android Police, Android Authority, Droid-Life, and XDA-Developers. He lives in Manhattan with his partner, a music therapist.
