On Tuesday, Google announced a new version of their main AI model, called Gemini 2.5 Pro Preview (I/O edition). Google says this new model does better on many common tests.

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

Google is releasing this model before their yearly I/O meeting for developers. At this meeting, they plan to show many new models, plus AI tools and platforms. Google is trying hard to get people to think about them and use their products in the very competitive world of AI. Other companies, like OpenAI and xAI, are also about to release models that people think will be very good.

Google says that Gemini 2.5 Pro Preview (I/O edition) is much better at coding and making interactive web apps. They also say the model is better at changing code to reach a specific goal and editing code.

In a blog post, Google says that Gemini 2.5 Pro Preview (I/O edition) is the best on the WebDev Arena Leaderboard. This is a test that checks how well a model can make 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 popular test called VideoMME.

Google wrote in a blog post that for developers already using Gemini 2.5 Pro, this new version will make coding better. It will also fix important problems based on what developers said, like making function calling have fewer errors and work more often. They also said the model naturally likes nice-looking web design but you can still control it.