Google's Gemini 2.5 Pro model is now available for free users: What's new

Released last week for paid subscribers, the Google Gemini 2.5 Pro model's experimental version brings advanced reasoning and coding capabilities

Google's new Gemini 2.5 AI model is now starting to roll out for free-tier Gemini users, starting with the experimental version of Gemini 2.5 Pro. Google said that the latest model can handle complex prompts and offers enhanced reasoning capabilities. The 2.5 Pro model is designed to analyse information deeply, draw logical conclusions, incorporate nuanced context and make informed decisions before responding.   

 

Last week, Google released the Gemini 2.5 Pro Experimental model to Gemini Advanced subscribers, but has now made it accessible to non-paying Gemini users as well. The latest AI model from Google is available for all through the Gemini mobile app, Gemini on web, and through Google AI studio. Despite being in its experimental phase, the model supports multiple features like file upload, extensions and apps.  

ALSO READ | China's Manus turns AI agent into subscription service with plans from $39 

What Gemini 2.5 Pro Model offers

 

Google refers to the Gemini 2.5 Pro Experimental as its most advanced model for handling complex functions. It is said to deliver strong results for reasoning and coding tasks. As per Google, the model leads in maths and science benchmarks such as GPQA and AIME 2025. It also scores 18.8 per cent across models without tool use on Humanity's Last Exam, a dataset developed to capture the human frontier of knowledge and reasoning.     ALSO READ | Google rolls-out custom chatbots 'Gems' for free-tier Gemini users: Details

 

As the first release in the Gemini 2.5 series, the 2.5 Pro model is said to be highly effective for tasks such as creating visually appealing web applications, building codes, and editing or transforming code. The Pro model leads many industry benchmarks by meaningful margins and demonstrates strong reasoning and coding abilities. In a demonstration video shared by Google, the model showcased its ability to generate a fully operational video game from a single-line prompt, producing executable code. The new model also offers users a glimpse into how it processes a request, and the steps it follows before delivering a response.