Google is forcing free Gemini users into auto mode and pushing more advanced models behind paid subscriptions to manage computing costs.
- Abolishing manual selection and enforcing auto mode on free accounts
- Introducing new effort and reasoning levels in the app
- Restructuring paid subscriptions and adjusting access privileges
- Motivations of cloud computing cost management and actual usage
- Frequently asked questions
Abolishing manual selection and enforcing auto mode on free accounts
Alphabet’s Google has halted manual selection of artificial intelligence models for users relying on the free version of the Gemini app without a paid financial subscription. Starting Friday, October 9, the company began mandatorily switching all free accounts to a specific setting called “Auto mode.” This new software mechanism automatically routes the majority of questions and inquiries to the “Flash Lite” model, which is the company’s lightweight and fastest model for processing simple commands.
This practical measure pushes Google’s more advanced and capable models behind paywalls and financial subscriptions. The specialized platform 9to5Google explains that the model selection menu for free accounts now shows only a single option labeled “Auto” with an explanatory phrase indicating “Gemini 3 models.” The company described this feature as relying on intelligent sorting to direct each prompt to its most appropriate model, confirming that routine prompts will go to Flash Lite, while requests requiring deep inference may be passed to the Flash or Pro models. If the user chooses to disable smart selection, all responses are generated exclusively by the Flash Lite model.
Guideline documents published by the company last week were clearer and more decisive, explicitly stating that free account holders will be restricted to using the Gemini 3.5 Flash Lite model and will lose direct access to the 3.6 Flash and 3.1 Pro models. Before this adjustment, any free user could freely switch between these three models, whereas anyone wishing to manually select the standard Flash model must now subscribe to the Google AI Plus plan, which costs $4.99 per month, according to The Verge.
Introducing new effort and reasoning levels in the app
Simultaneously with these changes, the company replaced the old “Expanded thinking” option with three new interactive effort levels to organize the model’s mental processing operations: Low, dedicated to “fast and high-efficiency” tasks; Medium, which achieves a “balanced depth”; and High, designed to deliver “highly accurate and comprehensive” analyses.
These new levels precisely match the “reasoning levels” already available to developers and programmers in the Google AI Studio and Antigravity environments. These options are currently being rolled out gradually to all subscribers, with a clear warning reported by The Next Web indicating that selecting high effort levels consumes a larger portion of the daily usage quotas available to each account.
Restructuring paid subscriptions and adjusting access privileges
The new adjustments affected current subscribers to the AI Plus plan, as Google announced that these users will lose access to the advanced Pro model, and their future access will be limited to the Flash Lite and Flash models only. Technical reports specified that the Flash model adopted in this plan will be version 3.8, although earlier company documents pointed to version 3.6. Subscribers are scheduled to receive emails explaining the timeline for this decision’s implementation.
Conversely, the company confirmed that subscribers to higher tiers—the AI Pro plan and the AI Ultra plan, which costs $19.99 per month—will not lose any of their current perks. Instead, they will receive an additional option called “Deep thinking,” dedicated to ultra-complex parallel logical computing. The company also revealed the upcoming arrival of a new model called “Argon” for Ultra plan subscribers, without disclosing specific details regarding the tiers it will fall under.
Motivations of cloud computing cost management and actual usage
These radical shifts come as part of a broader financial and technical strategy pursued by Google since last May to control escalating computing costs resulting from running generative artificial intelligence models across giant data centers. By directing routine requests to smaller, less energy-consuming models, the company aims to reduce operational burdens while prompting users with advanced needs to upgrade to regular monthly paid plans.
Tech industry analysts believe that restricting free accounts to smaller models reflects a new phase of monetizing artificial intelligence services, as tech giants can no longer provide advanced reasoning and logic capabilities without financial limits amid the skyrocketing costs of GPUs and the electrical power sources required to run them.
Frequently asked questions
Question: What is the new change Google imposed on free Gemini users?
Answer: The company abolished manual model selection and enforced an automatic selection system, routing most prompts and inquiries directly to the Flash Lite model.
Question: What are the new effort levels added to the app?
Answer: Google added three analysis levels: Low for speed, Medium for balanced depth, and High for comprehensive and deep analysis of issues.
Question: How were subscribers of the cheaper AI Plus plan affected?
Answer: Subscribers to this plan will lose access to the Pro model, and their usage will be restricted to the Flash Lite and Flash models only.
Question: What is the dedicated option for professional tier subscribers like Ultra?
Answer: Subscribers to the Pro and Ultra plans get the advanced Deep thinking feature, in addition to awaiting the arrival of the new model called Argon.
