Google will implement changes to access to Gemini artificial intelligence (AI) models, effective from October 9th. This modification will affect both users who use the service free of charge and those subscribed to the Google AI Plus plan.
According to information obtained by 9to5Google, based on a support document updated by Google, non-subscribers to an AI plan will lose the ability to access the Gemini 3.6 Flash and Gemini 3.1 Pro models. Their access will be restricted to the Gemini 3.5 Flash-Lite model.
This change follows an alteration made by Google in May, when the method for counting Gemini usage was adjusted to consider the amount of computational resources employed. Furthermore, the company will begin segmenting available models according to the user's contracted plan.
More details about the changes
Although free and AI Plus users will lose access to certain models, the AI Pro plan will receive a new feature. Subscribers to this plan, which costs US$ 19.99 monthly (approximately R$ 109), will have access to Deep Think, an option that Google defines as focused on 'maximum parallel reasoning.'
Previously, this function was exclusive to AI Ultra subscribers, whose plans are priced at US$ 99.99 and US$ 199.99 per month (equivalent to about R$ 544 and R$ 1,087).
Another update scheduled for October includes the introduction of three effort levels for models accessible in the application: low, medium, and high. The intention is to allow the user to define the degree of effort Gemini should dedicate to solving a specific task. According to the description cited by 9to5Google, the higher levels can enhance the model's ability to perform activities and generate more detailed responses, although this results in higher consumption of the usage limit.
Currently, Gemini already has the 'Extended thinking' option, intended for solving complex problems. The new proposed levels aim to align the application's operation with the functionalities already existing in Google AI Studio and Antigravity.
The exact location of Gemini 4 Argon, a new advanced Google model, within this plan structure has not yet been defined. Google communicated that the model will initially be released to AI Ultra subscribers, but did not specify whether Argon will be categorized as a Pro model or belong to a new superior category.
Google itself maintains support pages warning that usage limits and model availability are subject to change depending on variables such as system capacity and experimentation.
In practical summary, starting October 9th, the main transformation for those who use Gemini without payment will be the exclusion of access to the Flash and Pro models, and AI Plus subscribers will also stop using Pro. In return, the higher-tier plans will maintain access to the three models and gain new reasoning capabilities.