Google rolls out Gemini 2.5 Pro Experimental for free, but limits apply

In the United States on March 31, 2025, Google made its latest AI model, Gemini 2.5 Pro Experimental, freely accessible to users. This announcement came just days following its initial launch, which was exclusively for paying Google Gemini Advanced subscribers on March 25. The decision to open access was made to enhance user engagement by allowing more people to try the new AI features that Google claims lead in various benchmarks due to their strong reasoning and coding abilities. Gemini 2.5 Pro Experimental is reported to perform significantly better than its predecessor, Gemini 2.0, providing improved responses and code generation. While the free version offers access to most features of the new AI, it does not include all advanced functionalities available to paying users. Free users face several limitations, such as a lower maximum context window and rate limits that govern the number of requests they can make to the AI. The available context window for free users is smaller than the full capacity of 1 million tokens. In comparison, Gemini Advanced users enjoy a larger context window and higher request thresholds, indicating that Google is incentivizing subscriptions with superior access to its AI tools. The AI model's performance has been validated by comparisons with competitors, such as Grok 3 Preview and ChatGPT 4.5 Preview. Google emphasized that Gemini 2.5 Pro Experimental scored high on the LMArena leaderboard, securing favorable ratings from human evaluators, highlighting its competitive edge in AI technologies. Despite the positive reception, the limited access for free users raises questions among existing subscribers about the necessity of maintaining their paid subscriptions. Google has hinted at potential changes to the token and usage limits as they gauge user feedback during this rollout period. Google’s swift move to provide free access seems to reflect an increasing urgency to compete with rivals like OpenAI, who have been pivotal in advancing AI accessibility. Furthermore, Google's plans to enhance the model’s capabilities by expanding the context window from 1 million to 2 million tokens signals a commitment to further improving AI tools. The expectation is that this could eventually lead to a more streamlined version of Gemini for everyday tasks, potentially reshaping how users interact with AI technology.