Google Tightens Access to Free Gemini Models
Google updated the Gemini application support page, stating that starting October 9, personal accounts without an AI subscription can only use Flash-Lite. The previously available Flash and Pro for free users will no longer be offered. This change will first take effect for non-subscribers.
Google AI Plus retains Flash-Lite and Flash but loses Pro. The effective date will be individually notified to accounts via email. According to 9to5Google, the monthly fee for this tier is $4.99. Google AI Pro and AI Ultra retain all three tiers: Flash-Lite, Flash, and Pro. The Pro monthly fee is $19.99, and it adds Deep Think, which was previously only available in Ultra. The monthly fees for Ultra are $99.99 and $199.99.
Usage multipliers are listed on the same support page. Non-subscribers have the standard quota. AI Plus has double the standard quota. AI Pro has four times the standard quota. AI Ultra varies by tier, offering five or twenty times that of AI Pro. Media generation, Deep Research, Pro models, extended reasoning, and Deep Think consume more quota.
9to5Google lists the models as 3.5 Flash-Lite, 3.6 Flash, and 3.1 Pro. The October update will also introduce low, medium, and high effort levels for each available model. Higher tiers will result in more complete task completion and responses, consuming more quota. The page specifies personal accounts without listing country exceptions.
No new models were released this time. The change is about who can access which tier. The free tier has been reduced from three tiers to one, while the $4.99 tier has shifted from access to Pro to two lightweight models. Full models and parallel reasoning remain at $19.99 and above.
In market mechanics, the buyers are Gemini individual users, and the sellers are Google's reasoning quotas. The event is driven by the closure of standard Flash for the free tier, not the launch of new subscriptions. Usage is shifting from the zero-cost Flash to paid tiers. Beneficiaries are the Pro and Ultra tiers at $19.99 and above, while those relying on free Flash for daily tasks are under pressure.
Source: Public Information
ABAB AI Insight
Google already changed the Gemini usage to be calculated by computational load in May, and this October change is about model eligibility. The free tier retains the lightest Flash-Lite, the $4.99 Plus retains Flash, and the Pro model is now at $19.99. Deep Think has been moved down from $99.99 and $199.99 to $19.99, effectively justifying a mid-tier with a high-priced feature.
No new funding has been introduced. Resources are being withdrawn from free reasoning quotas to be allocated to paid tiers. The motivation is that the single-use costs of Flash and Pro are higher than Flash-Lite, and free access essentially subsidizes reasoning with advertising and search profits. The support page lists high effort, deep research, and media generation as more quota-consuming, tying usage and models to the same price list.
This mirrors OpenAI's strategy of keeping free users on lightweight models while placing reasoning models in Plus and Pro. It is also similar to cloud providers removing GPU instances from free trials. The industry positioning is subscription tiered, not model competition. The 3.1 Pro has not been replaced; only the eligible users have changed.
Structurally, this represents a transfer of pricing power. Model capabilities are no longer determined by opening the application but by the monthly fee tier. The reason this occurs is that reasoning has marginal costs; as usage in the free tier increases, standard models will directly consume quotas. Removing Flash from the free tier means the conversion happens the day users realize the original models are no longer available.
ABAB News · Cognitive Law
The free tier retains the cheapest option.
The models have not changed; the eligible users have.
The quota multipliers are a price list, not technical parameters.