Google suddenly offers free 'super AI' to all users

This surprising move comes just days after Google launched Gemini 2.5 Pro Experimental for users who subscribe to the Advanced paid plan.

Last week, Google introduced Gemini 2.5 — the company’s most advanced AI model to date, with built-in reasoning capabilities. The model can analyze complex information with contextual understanding to draw logically accurate conclusions.

Google suddenly gives Gemini 2.5 Pro for free to all users. Photo: Shutterstock

This new model was initially only available to those willing to pay $20/month for the Gemini Advanced package. However, less than a week later, Google surprised everyone by opening up free access.

“The team is working at full capacity, the TPU is running at full capacity, and we want to get our smartest model into the hands of more people as soon as possible,” Gemini App’s official X account wrote in a series of posts explaining the decision.

With many Advanced subscribers responding to the question of why they should pay if free users can also use the new model, the official account emphasized that Advanced users still have a better experience. Free users are limited in frequency of use, while this does not apply to Advanced users.

The official documentation also confirms that advanced users will still want to sign up for a paid account. Specifically, for the Gemini 2.5 Pro Experimental model, Advanced users can make 4 times more requests per minute (20 vs. 5), 4 times more requests per day (100 vs. 25), and double the number of tokens per minute (2 million vs. 1 million).

Despite being released less than a week ago, Gemini 2.5 Pro has already made a strong impression. The model is currently at the top of the LMArena leaderboard, surpassing ChatGPT-4o and beating its competitors in math, creative writing, and science . However, it still lags behind OpenAI’s model in programming and multi-turn conversation handling.

Notably, Gemini 2.5 scored 18.8% on Humanity’s Last Exam, a test that measures an AI’s ability to use complex knowledge. This achievement puts it ahead of rivals like OpenAI o3-mini (14%), GPT-4.5 (6.4%), Claude 3.7 Sonnet (8.9%), and DeepSeek R1 (8.6%).

Still, the Gemini 2.5 Pro falls short of OpenAI's Deep Research model's 26% score. However, the comparison isn't entirely fair, as Deep Research has the ability to search the web, giving it an advantage on questions that require general knowledge.

Source: https://vietnamnet.vn/google-bat-ngo-mien-phi-sieu-ai-cho-tat-ca-nguoi-dung-2386109.html