Google Limits Free Gemini to Flash-Lite on October 9

Free Gemini users will soon lose access to two model classes they can currently use. Google says that beginning October 9, 2026, personal accounts without an AI subscription will have access only to Flash-Lite in Gemini Apps. Flash and Pro will no longer be available.

The entry-level subscription won’t preserve the full selection either. Google AI Plus will retain Flash-Lite and Flash but lose Pro, on a schedule Google will communicate to subscribers by email. AI Pro and AI Ultra keep all three model classes, while AI Pro also gains Deep Think access.

Anyone relying on Flash or Pro in the Gemini app now faces a decision: finish time-sensitive work before access starts changing, or consider a subscription that retains the required model. The announcement does not establish an equivalent October 9 cutoff for the Gemini API or Google AI Studio, which have separate access and billing arrangements.

October 9 Is the Deadline to Plan Around

Google’s model access notice applies to Gemini Apps used with a personal account. For users without an AI subscription, the changes will “start to take effect” on October 9.

October 9 marks the beginning of the change, with no promise that every account will switch at the same moment. Google provides no exact cutoff time or time zone in the supplied notice. Free users should plan around losing Flash and Pro access from that date instead of counting on a later rollout.

The support page is undated. The accompanying 9to5Google report was published on October 3; that reporting date should not be confused with a timestamped Google announcement.

Keeping a conversation does not necessarily mean retaining the model used to produce its answers. The notice announces a model-access change and says nothing about deleting existing chats. It also leaves open whether an existing conversation can continue with its original model after the account loses access. Users should not assume an open thread will preserve that entitlement.

The New Gemini Access Matrix

Once the changes take effect for the relevant account, Google’s plan matrix is:

Free users will lose model choice across those three classes. This goes beyond a lower allowance for Flash or Pro: neither model will be included in free Gemini Apps access.

AI Plus preserves Flash but will no longer provide Pro. Subscribers who chose Plus for occasional Pro use will need to reassess the plan, even if its remaining usage allowance is sufficient for their workload.

The October 9 date is explicitly attached to users without a subscription. AI Plus subscribers should expect an email explaining when their access will change. The notice gives no basis for treating October 9 as a universal Plus cutoff.

The October 3 report lists AI Plus at $4.99 per month and AI Pro at $19.99 per month. These are reported subscription prices, not Gemini API rates. Readers should check the offer displayed for their own account before upgrading. The documented change concerns what the plans include; no increase in subscription prices has been announced.

AI Pro is the lowest listed Gemini Apps subscription that retains Pro-model access. Buying AI Plus to recover Flash would not also restore Pro.

AI Pro Gains Deep Think, but Effort Still Uses Quota

AI Pro gains a capability in this change. Google says AI Pro and AI Ultra subscribers will have access to Deep Think, which it describes as an option for “maximum parallel reasoning.” The accompanying reporting identifies this as an expansion to AI Pro from the higher Ultra tiers.

Users will also be able to choose low, medium, or high effort for each model available to their plan. Google says increasing effort can improve a model’s ability to complete tasks and produce more thorough answers, while consuming more of the user’s limit.

Image: Gemini app limiting what models free & AI Plus users can access, AI Pro adding Deep Think.

These are two separate controls:

• Model access determines whether an account can use Flash-Lite, Flash, or Pro.
• Effort level determines how much reasoning effort an available model applies to a request.

High effort does not unlock a model excluded from the subscription. A free account using Flash-Lite at high effort would still be using Flash-Lite, without gaining Pro through that setting. Google’s notice provides no benchmark evidence establishing that higher effort makes one model equivalent to another.

Deep Think is an included capability with usage limits. Google’s support page describes compute-based limits that refresh every five hours until the user reaches a weekly limit. Prompt complexity, the features used, and conversation length all contribute to consumption.

Under that framework, the page lists AI Plus at twice the standard allowance and AI Pro at four times the standard allowance. Those multipliers describe usage limits, not model eligibility. Remaining quota on Plus will not restore Pro after its access change takes effect.

Gemini Apps Restrictions Are Not an API Cutoff

“Gemini access” can refer to several services. The October notice concerns Gemini Apps with personal accounts. It does not declare that every free developer account, API key, or AI Studio session will lose Flash and Pro on October 9.

Google’s AI Studio subscription documentation describes a separate system. Google AI Pro and Ultra subscriptions provide additional model access and higher quotas within the AI Studio web interface, including its Playground and Build interfaces. The documentation also lists a free AI Studio tier with basic access and a modest quota.

Those subscription benefits apply within the AI Studio interface. Direct Gemini API use through API keys or external applications is billed and managed separately.

Readers should distinguish three situations:

• Using Gemini Apps: The October plan matrix determines access to the listed model classes.
• Working inside AI Studio: Consult AI Studio’s model availability and quotas, including any Pro or Ultra subscription benefits.
• Calling the Gemini API: Check the API project’s model access, usage tier, and billing configuration rather than assuming a consumer subscription covers requests.

Google’s developer documentation describes continuing work through a Cloud Billing-enabled Gemini API key on a pay-per-request basis after AI Studio subscription quotas are exhausted. It also says AI Studio’s daily subscription allowances use resets rather than rolling time windows.

The quota systems differ even when the same Google account is involved. An AI Pro subscription may provide benefits in both Gemini Apps and AI Studio, but it is not a blanket payment for API calls.

The supplied announcement establishes no identical model restriction in AI Studio or the API. Developer access can still change, so developers need to check the documentation governing their particular service.

What to Check Before Your Access Changes

If your work in free Gemini Apps depends specifically on Flash or Pro, plan around October 9. Save important outputs and complete deadline-sensitive tasks before the change begins. An existing conversation should not be assumed to retain the same model.

For AI Plus subscribers, the next step is to watch for Google’s email. Flash remains included, so an upgrade is only necessary to preserve access to something the plan is losing, such as Pro, or to obtain additional benefits.

AI Pro subscribers should weigh Deep Think against its usage cost. More reasoning options may be useful, but Google has supplied no comparative results in this notice showing how much improvement users should expect for their particular tasks.

One model-roadmap question remains unresolved: the October 3 reporting says Google has not explained where Gemini 4 Argon fits within the plan matrix. Retaining access to the Pro class should not be treated as a promise of access to every future Gemini model.

Base the immediate subscription choice on the published entitlements. Flash-Lite remains free, Flash requires at least AI Plus under the new matrix, and Pro requires at least AI Pro. For affected users, the subscription tier will determine model choice more directly, regardless of how much unused quota remains.