Google Overhauls Gemini App Model Tiers and Introduces New Effort Levels for Subscribers

Google is restructuring access to its artificial intelligence models across its subscription tiers, introducing tighter boundaries on which versions of Gemini are available to free users and entry-level paid subscribers. Building upon the compute-based usage changes implemented earlier this year in May, the tech giant is rolling out a revised framework that clearly delineates model availability according to user plans, while simultaneously expanding advanced reasoning capabilities for its highest-paying customers.

According to a newly updated support document published by Google, Gemini app users who operate without a paid Google AI subscription will soon face stricter limitations regarding the underlying technology powering their interactions. Beginning on October 9, individuals using the free tier of the Gemini app will be restricted exclusively to the 3.5 Flash-Lite model. Consequently, free users will lose access to more capable architectures, including the 3.6 Flash and 3.1 Pro models, which have previously been accessible under varying conditions.

The adjustments also impact the entry-level paid tier known as AI Plus. Subscribers to this $4.99-per-month tier will find their model choices capped at Flash-Lite and Flash, with the Pro model being removed from their available options. Google has indicated that customers currently subscribed to the AI Plus tier will receive direct email notifications outlining the precise timeline for when these structural changes will take effect on their accounts.

In contrast, users subscribed to the higher-end AI Pro and AI Ultra tiers will experience no regressions in their current model access. For the $19.99-per-month AI Pro customers, the update actually brings a notable enhancement. These users will gain access to the Deep Think option, a feature designed for maximum parallel reasoning that was previously restricted to the company’s most expensive plans, which cost $99.99 and $199.99 per month. This expansion ensures that mid-to-high-tier professional users can leverage advanced computational problem-solving without necessarily jumping to enterprise-level pricing brackets.

Alongside these tier-based adjustments, the Gemini app is slated to receive granular controls over how the AI processes queries. Sometime in October, Google plans to introduce new low, medium, and high effort levels for each available model within the application. According to company documentation, these adjustable settings will increase a model’s capacity to complete complex tasks and deliver significantly more thorough, detailed answers to user queries. However, this deeper processing comes with a trade-off, as higher effort settings will consume a larger portion of the user’s allocated usage limits.

This upcoming implementation mirrors the thinking levels already integrated into Google AI Studio and Antigravity environments, marking a transition of advanced developer-facing controls into consumer-facing applications. Currently, Gemini app users are limited to enabling a broader setting known as extended thinking for complex problem solving, making the arrival of tripartite effort levels a more flexible approach for everyday interactions.

Amid these updates, questions remain regarding the positioning of upcoming proprietary architectures. Google recently announced that its Gemini 4 Argon model will debut exclusively for AI Ultra subscribers. However, the company has not yet clarified whether Argon will be formally classified under the Pro model family or if it will anchor an entirely new tier situated above the existing offerings. As Google continues to refine its AI ecosystem, these tiered rollouts underscore the delicate balance between delivering cutting-edge computational power and managing the substantial server resources required to sustain global demand.