Google disclosed on Tuesday the introduction of Gemini 2.5 Pro Preview (I/O edition), an enhanced iteration of its core Gemini 2.5 Pro AI model, which the company posits surpasses various established benchmarks.

You can use the Gemini 2.5 Pro Preview (I/O edition) through the Gemini API, Google’s Vertex AI, and AI Studio. It costs the same as the Gemini 2.5 Pro model it replaces. You can also find it in Google’s Gemini chatbot app for the web and mobile phones.

The model's debut precedes Google's annual I/O developer conference, accounting for its 'I/O edition' moniker. At this event, Google is anticipated to unveil a plethora of models, alongside AI-driven tools and platforms. The corporation is locked in intense competition for both mindshare and marketshare within the fiercely contested AI arena; competitors such as OpenAI and xAI are poised to launch models projected to exhibit exceptional performance.

Google says that Gemini 2.5 Pro Preview (I/O edition) is much better at coding and creating interactive web apps. The company also says the model is better at changing code to do something specific and editing code.

Google announced in a blog post that Gemini 2.5 Pro Preview (I/O edition) holds the top position on the WebDev Arena Leaderboard, an evaluation tool that assesses a model’s proficiency in developing visually appealing and practical web applications. The post also highlighted Gemini 2.5 Pro Preview (I/O edition)'s cutting-edge capability in video comprehension, evidenced by its score of 84.8% on the widely recognized VideoMME benchmark.

In its blog post, Google stated that for developers already utilising Gemini 2.5 Pro, this latest iteration will not only enhance coding efficiency but will also tackle significant developer concerns, such as decreasing function calling errors and increasing function calling activation rates. They added that, by default, the model exhibits a strong preference for visually appealing web development while retaining its controllability.