
Free users are also restricted from using the Flash model; Pro is available for plans of $19.99 or more.
Google has revamped its AI Gemini plan to block access to the 'Pro' model for free and low-cost subscribers, and is seeking to attract subscribers to the high-priced plan ahead of the launch of its new top model, 'Gemini 4 Argon'.
AI-generated summary
After releasing the 'Gemini 3.1 Pro' last February, Google introduced the top model 'Gemini 4 Argon' on the 30th of last month.
(San Francisco = Yonhap News) Correspondent Kwon Young-jeon = Google revamped its artificial intelligence (AI) model Gemini plan, preventing low-cost subscribers and free users from accessing the 'Pro' model.
According to a Google Customer Center announcement on the 4th (local time), starting this month, Google will lower the level of AI models that can be accessed by plan.
The Gemini Pro model, which was previously available to some free users, will be available starting with the 'AI Pro' plan with a monthly subscription fee of $19.99 (KRW 29,000 in Korea).
Accordingly, not only free users but also subscribers of the low-cost 'AI Plus' plan, which costs $9.99 per month (KRW 7,500 in Korea), will no longer be able to use the pro model.
Google has revised its rate plan so that free users cannot use the lightweight model 'Gemini Flash' and can only use the 'Gemini Flash-Lite' model, which has lower performance.
Google plans to restrict free users' access to Gemini Flash and Pro starting on the 9th, and subscribers will be individually notified by email when specific plan changes will be applied.
However, Google allows users to select the model's inference level from 'low', 'medium', and 'high'.
In general, even models with low performance can provide higher quality answers by increasing the level of inference, but in this case, the AI usage limit granted by Google will also be used more.
AI Pro and AI Ultra plan users with a monthly subscription of $99.99 or more can also use the 'Deep Sync' function, which uses multiple models at the same time to make inferences.
Google currently uses an AI usage-based cap system that resets every five hours.
Google's plan reform is interpreted as a rate increase ahead of the official launch of the new top model.
After releasing the 'Gemini 3.1 Pro' in February, Google's pro model development was sluggish for about 7 months, and it only introduced the top model 'Gemini 4 Argon' on the 30th of last month.
Accordingly, it is believed to be a strategy to attract subscribers to high-priced plans by utilizing the Gemini 4 Argon model, which consumes large computing resources.
However, in this process, voices of dissatisfaction are expected to arise as free users are prevented from using the pro model as well as the lightweight model.
AI outlook — possibilities, not facts
Google plans to restrict access to Gemini Flash and Pro models for free users from the 9th
Very likely · Within days

Recently, personal information leaks have continued from banks, platforms, and public institutions, and concerns have been raised that leaked income, loans, workplaces, contact information, etc. are being combined with generative AI to create precise personal profiles and then used for spear phishing. Targeted phishing attacks using AI have already been reported overseas, and cases where the click-through rate was 4.5 times higher than the existing method were observed.

Avery Ching, co-founder and CEO of Aptos Labs, predicted at a press conference in Seoul that future finance will become an on-chain blockchain economy, and said that Aptos viewed regulations as design requirements and built an infrastructure that prioritizes safety and security. He also emphasized the possibility of combining stablecoins and AI in the Korean market and said that he is planning to expand tokenization asset pilot projects through collaboration with institutions.

US President Donald Trump named artificial intelligence (AI) 'super intelligence (SI)' and launched a White House task force dedicated to this. Director of National Intelligence Jay Clayton has been appointed AI czar, tasked with coordinating cooperation between the government and private companies and securing America's technological leadership.

State-run Korea Electric Power Corp. announced that personal data of roughly 24,000 employees was leaked on an external website. The exposed data includes names, titles, and phone numbers, though sensitive information remained safe.

The Ministry of Science and ICT has launched a 24-hour emergency response system with the Korea Internet & Security Agency (KISA) in relation to the recent financial hacking incident and has begun to prevent the spread of the virus to the private sector and further damage.

From January to August of this year, there were approximately 240,000 hacking attempts targeting the court computer network, significantly exceeding the number of hacking attempts for the entire year last year. In addition, cases of personal information leakage due to mistakes by internal employees or AI-related factors were found to have increased significantly compared to previous years.