Google has announced the launch of an experimental version of the Gemini 2.5 Pro neural network for all users of the Gemini app, including free accounts. This was an unexpected move, as the algorithm was announced just a few days ago as the company’s “most intelligent artificial intelligence model”, capable of not only answering questions but also reasoning.
Currently, Gemini 2.5 Pro still has the status of an experimental model, but supports a number of features, including expansion, file uploading and Canvas. The algorithm was initially only available to Gemini Advanced subscribers, but over the weekend Google announced plans to open up access to all users.
You can interact with Gemini 2.5 Pro now at gemini.google.com, and the model will soon be available in the Gemini apps for Android and iOS. Google says it is committed to making its advanced AI model available to a wider audience.
Note that as of today, Gemini 2.5 Pro tops the LMArena rankings, which rates models based on user preferences. The algorithm also leads in mathematical (AIME 2025) and scientific (GPQA diamond) tests. In addition, Google is actively improving the neural network’s programming capabilities.
Whether the stable version of Gemini 2.5 Pro will become free for all users is still unknown. Perhaps the current expansion of access is part of testing. Google has not made any official statements on this issue yet.