On Tuesday, Google unveiled the Gemini 2.5 Pro Preview (I/O edition), an enhanced iteration of its leading Gemini 2.5 Pro AI model, which the company asserts surpasses several prevalent benchmarks.

The Gemini 2.5 Pro Preview (I/O edition) is accessible through the Gemini API and Google's Vertex AI and AI Studio platforms, carrying the same price as the Gemini 2.5 Pro model it largely supersedes, and is also integrated into Google's Gemini chatbot application for both web and mobile use.

The model is being released before Google's yearly I/O developer conference (which is why it's called the "I/O edition"). At this conference, Google is expected to show many new models, as well as AI tools and platforms. The company is trying hard to be the most popular and successful in the very competitive AI market. Other companies like OpenAI and xAI are also about to release models that are expected to work very well.

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

In a blog post, Google says that Gemini 2.5 Pro Preview (I/O edition) is number one on the WebDev Arena Leaderboard. This test measures how well a model can create good-looking and useful web apps. Gemini 2.5 Pro Preview (I/O edition) is also very good at understanding videos, getting a score of 84.8% on a well-known test called VideoMME.

Google stated in its blog post, “For developers already using Gemini 2.5 Pro, this new version is poised to enhance coding performance and resolve critical developer concerns, such as minimising errors in function calling and boosting function calling trigger rates.” They added, “By default, the model exhibits a distinct preference for aesthetic web development, all while retaining its capacity for guidance.”