Google has taken the free Gemini experience down to its simplest form. Since October 9, 2026, Ground News reports, anyone using Gemini without a paid subscription has been limited to a single model, Gemini 3.5 Flash-Lite, cut off from the heavier Gemini 3.6 Flash and Gemini 3.1 Pro models that free users previously got limited access to, a reduction from three free models to just one. Google AI Plus subscribers are a separate case: Google told them on October 16 that selectable Pro models will leave that plan on November 13, 2026. Meanwhile, on October 13, 2026 Gemini 4 Argon expanded beyond its initial cybersecurity-defender cohort to paying subscribers and businesses, pushing the frontier even further out of free users’ reach.
This is not a quiet quota tweak. It was a structural narrowing of what “free Gemini” means, and it landed five months after Google already rewired how usage gets measured across the Gemini app, and only weeks after Google kept expanding the Flash lineup itself, shipping Gemini 3.7 Flash on August 13, 2026 and Gemini 3.8 Flash on September 2, 2026, both with a 1,048,576-token context according to Promptzone. Together, the two moves point at a company trying to push casual users toward paid tiers while reserving its best models, and its heaviest compute, for subscribers.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Changed in the Gemini App on October 9
Before the change, a free Gemini account, which Google lists at $0 a month with a Google Account and which still includes Flash-Lite, could select Flash-Lite, Flash, and a capped amount of Pro inside the Gemini app. According to 9to5Google’s October 3 report and Google’s own support documentation, free accounts dropped to Flash-Lite only as of October 9: Gemini 3.6 Flash and Gemini 3.1 Pro were removed from the model picker entirely. Digital Trends and The Economic Times both independently reported the same cutoff date and the same outcome, describing it as free users losing both of the stronger models rather than simply hitting lower caps on them.
The distinction matters. A quota cut still lets you touch the good model occasionally. A model-eligibility restriction removes the choice altogether. Reports citing Google’s support pages are explicit that this was framed as an eligibility change, not a shutdown of Gemini Pro as a product. Paying tiers keep full access; free tiers are now boxed into the entry-level model.
Flash-Lite is not a toy model, but it is built for speed and low compute cost rather than depth. It handles quick lookups, short drafts, and simple reasoning well. It is not the model Google points to for long documents, multi-step coding tasks, or anything requiring the extended reasoning that Pro and the Deep Think feature are designed for. For the millions of people who use Gemini casually through Google Search’s AI Mode or the standalone app, the day-to-day experience may not feel dramatically different. For anyone who occasionally leaned on Pro for a harder task, that option is now gone unless they pay.
Gemini Access by Plan, Before and After October 9
| Plan | Monthly price | Models before Oct 9 | Models after Oct 9 |
|---|---|---|---|
| No subscription (free) | $0 | Flash-Lite, Flash, limited Pro | Flash-Lite only |
| Google AI Plus | $4.99 | Flash-Lite, Flash, Pro | Flash-Lite, Flash, Pro (Pro removal scheduled for November 13, 2026) |
| Google AI Pro | $19.99 | Flash-Lite, Flash, Pro | Flash-Lite, Flash, Pro, plus Deep Think |
| Google AI Ultra | $99.99 | Full model set, highest limits | Full model set, highest limits, unchanged |
Pricing for the three paid tiers, $4.99 for Plus, $19.99 for Pro, and $99.99 for Ultra (Google also lists a higher Ultra tier at $199.99 a month), comes from reporting referenced in the October coverage of the change. What stands out in the table is that AI Plus subscribers, people already paying Google roughly five dollars a month, are set to lose Pro access too: Google announced on October 16 that selectable Pro models will leave the plan on November 13, 2026. Only Google AI Pro and Ultra subscribers will keep the full lineup, which makes Pro the plan to move to, a mid-tier jump from $4.99 to $19.99. References to “Flash” and “Pro” throughout this article mean Gemini 3.6 Flash and Gemini 3.1 Pro, the two models Google removed from the free tier on October 9.
The New Effort-Level System: Low, Medium, High
Alongside the model restriction, AI Weekly’s October 4 report describes Google introducing selectable “low,” “medium,” and “high” effort levels inside Gemini, with Ground News reporting that Google added these effort controls specifically to free-tier models in October 2026. Ground News also reported that higher effort settings burn through usage limits faster than lower ones, confirming that effort level is not just a quality dial but a direct lever on how quickly someone exhausts their five-hour or weekly compute window. That gives Google a second axis of control sitting on top of the model restriction itself: not just which model you can reach, but how much reasoning effort that model is allowed to spend on a given answer, and how fast that effort draws down your remaining quota.
This mirrors a pattern already visible elsewhere in the industry, where providers decouple “model” from “how hard the model tries” so they can sell compute in finer slices. With higher effort levels now confirmed to drain usage limits faster, free and lower-tier users have a practical reason to default to “low” for routine tasks and save “high” for questions that genuinely need the deeper reasoning. If effort levels end up gated by subscription tier the way model access already is, Google would have two separate levers to push free users toward a paid plan instead of one.
Why October 9? The Compute Math Behind the Cutback
The October 9 change did not happen in isolation. It follows a quieter shift that reporting traces back to May 2026, when Google moved Gemini’s usage accounting away from simple prompt counts and toward compute-based limits, with a usage window that resets every five hours plus an additional weekly cap. That overhaul is reportedly still in place and running underneath the new model restrictions.
Under that system, the paid tiers carry relative usage multipliers against the free baseline: AI Plus at 2x and AI Pro at 4x, while AI Ultra offers 5x higher usage limits vs. AI Pro at $99.99 a month or 20x higher usage limits vs. AI Pro at $199.99 a month. Pro and Deep Think are expensive to run at scale. Giving free users unrestricted shots at either one, even occasionally, eats into the exact compute budget Google is trying to protect. Pulling Pro out of the free tier entirely, rather than just capping it harder, is the cleaner way to guarantee that budget stays intact.
Enterprise and Education Accounts Are Treated Separately
One detail worth flagging for anyone using Gemini through a school or workplace account: the restrictions described in this round of reporting apply to personal Gemini accounts. Organizational and enterprise plans are reportedly tiered separately, meaning a business or education deployment of Gemini is not necessarily subject to the same October 9 cutoff as a personal Google account. Google has not published a single unified page covering every account type’s eligibility, which is part of why the rollout has generated so much confusion.
How Gemini’s Free Tier Now Compares to ChatGPT, Copilot, and Claude
None of Google’s three biggest AI rivals have announced an equivalent move in the weeks since Google’s announcement, which makes the comparison worth laying out plainly. OpenAI’s ChatGPT free tier gives limited access to its more capable models and tools, with exact caps shifting by model, demand, and feature rather than a hard model lockout. Microsoft’s Copilot remains free to use across web and app, with availability varying by product and region rather than a flat tier wall. Anthropic’s Claude offers free access with usage capped over a rolling period, detailed on Anthropic’s pricing page, where higher caps and priority access are reserved for paid plans.
| Service | Free-tier model access | Primary limitation |
|---|---|---|
| Google Gemini | Flash-Lite only, since Oct 9, 2026 | Flash and Pro require a paid plan; usage also capped on a 5-hour / weekly compute window |
| OpenAI ChatGPT | Limited access to higher-capability models | Caps vary by model, demand, and feature rather than a flat model lockout |
| Microsoft Copilot | Free web and app access | Feature and limit availability varies by product, account, and region |
| Anthropic Claude | Free access with rolling usage caps | Caps reset over a rolling period; paid plans raise limits and add priority |
What this table makes clear is that Google’s move is more blunt than how its rivals currently handle free access. Removing model choice outright, rather than tightening a quota, is a sharper line than ChatGPT, Copilot, or Claude currently draw for their free users. Whether that is a deliberate differentiator or simply what the compute math forced Google into, it puts Gemini’s free tier at the strict end of the current field.
Does This Touch the Gemini API or Just the Consumer App?
The reporting behind this change is specifically about the consumer-facing Gemini app and the web client tied to personal Google accounts, not the developer-facing Gemini API billed through Google AI Studio or Vertex AI. Those surfaces already run on their own pay-as-you-go pricing, separate from the AI Plus, Pro, and Ultra subscription ladder, and nothing in the current coverage suggests API pricing or free-tier API quotas changed alongside the October 9 rollout. Google’s API free tier includes 5,000 Search-grounding requests a month, APIDog reports, and Search grounding costs $14 per 1,000 requests after the free allowance.
That distinction matters for developers who have been testing Gemini models for side projects or production workloads through the API rather than the consumer app. For now, the restriction described in this round of reporting is about what a logged-in Gemini.google.com or mobile-app user can select from the model dropdown, not about API access tokens or billing tiers. For a sense of how API pricing sits across the Flash line, CloudZero reports that Gemini 3.5 Flash launched on May 19, 2026 at $1.50/$9 per million tokens. It is worth watching whether Google eventually brings similar model-eligibility logic to free API tiers, but that is a separate decision the company has not signaled yet.
Where Gemini 4 Argon Sits Relative to the Free Tier
Google released Gemini 4 Argon on October 13, 2026, its newest publicly announced frontier model. Google says it is available through the API, AI Studio, the Gemini platform, Google Antigravity, the Gemini app, and AI Mode in Search, with access depending on the user’s plan. The announcement ties consumer access in the Gemini app and AI Mode to Google AI Pro or Ultra subscriptions. Gemini 4 Argon is not documented as generally free in Google’s announcement. For someone on the free tier, that means the newest model sits behind the same paywall as the models removed on October 9.
Developers see a different price list. Google lists an introductory price of $2 per million input tokens and $10 per million output tokens (with cached input tokens at 95% off). After the introductory period expires on November 30, 2026, the price of $4 per 1M input tokens and $20 per 1M output tokens will apply. Anyone planning an Argon project on the API should budget for the higher figure rather than the introductory one.
Google’s own page says Argon ties for #1 on CWE-Bench (68%) for vulnerability remediation, and that it is expanding to hundreds of additional Fairwind Program partners. Google engineers have applied Argon to everyday work including debugging, large-scale codebase migration, and algorithm design, and on DeepSWE v1.1, a benchmark of real-world long-horizon software engineering tasks, Argon scored 77.9%. None of that reaches the free plan, which stays at $0 / month with a Google Account and 15 GB of cloud storage across Gmail, Drive, and Photos. Readers who only need quick answers can stay on Flash-Lite, and those who want Argon should compare the plan pages before paying.
Market Impact: A Push Toward Subscription-First AI
Every major AI lab is running the same equation right now: inference is expensive, free users rarely convert on their own, and the fastest way to change that math is to make the free experience noticeably worse than the paid one. Google has spent the better part of 2026 giving away heavyweight models to build Gemini’s user base, which surpassed 1 billion monthly users in August 2026 according to Sundar Pichai as relayed by AIBusinessWeekly, and prove out its consumer AI platform. October 9 marked the point where that strategy flipped toward extraction.
The $4.99 AI Plus tier is the interesting pressure point here. Stripping Pro out of a plan that people are already paying for, which Google has scheduled for November 13, 2026, is a way of nudging existing subscribers up to the $19.99 Pro tier rather than just pushing free users toward Plus. That is a more aggressive upsell structure than a typical freemium ladder, and it suggests Google sees meaningful revenue upside in moving its mid-tier base up a rung, not just in converting non-payers.
There is also a compute-allocation story sitting underneath the pricing story. Running Pro and Deep Think at scale for anyone who asks, paying or not, means provisioning capacity for the heaviest possible demand curve. By narrowing exactly who can even request those models, Google gets a more predictable load profile to plan data center and chip capacity against, which matters at a moment when GPU and accelerator supply across the industry remains tight. A tier wall is, among other things, a capacity-planning tool.
Historical Context: From Bard’s Free-for-All to a Tiered Gemini
Google’s conversational AI started life as Bard, a free, unrestricted chatbot the company rolled out in 2023 as a response to the ChatGPT boom. The rebrand to Gemini folded that free-for-everyone posture into a broader family of models spanning Flash-Lite up through the heaviest Pro variants, all offered with generous, loosely metered free access for most of Gemini’s early life.
May 2026’s switch to compute-based accounting was the first real sign that era was ending. October 9’s model lockout was the second, and the more visible one, because it changed what free users could see and click, not just an invisible counter behind the scenes. Four days later, on October 13, 2026, Google’s own API release notes confirmed a third step: the release of Gemini 4 Argon to paying subscribers and businesses, which raised the ceiling on the company’s top-end capability at the very moment free users were being boxed into the entry-level model. The trajectory across 2026 has been consistent: looser free access early in the year, tighter compute accounting by May, an outright model restriction by October, and now a widening gap at the frontier itself. Each step has reduced what a non-paying user can do even as Google keeps pushing what it advertises as Gemini’s top-end capability further out.
Industry Reaction
Coverage of the change has leaned critical. Digital Trends characterized the shift as a “major downgrade” for free users because it removes model choice rather than simply trimming a quota. Other secondary coverage frames it as Google drawing a clearer line between free, low-cost, and premium tiers, and as a monetization and compute-allocation decision meant to reserve its strongest models for paying and enterprise customers. None of the reporting identifies a named industry analyst offering on-record commentary, so the criticism so far is coming from the outlets covering the change rather than from a quoted third party.
What Free Gemini Users Should Do Now
The October 9 cutoff has already passed, so anyone relying on Gemini Pro for free is no longer waiting for a change, they are already living with it. A few practical points worth knowing:
- Check which model you are currently set to inside the Gemini app’s model picker; Gemini 3.6 Flash and Gemini 3.1 Pro have disappeared from that menu for free accounts now that the change has rolled out.
- If a specific task genuinely needs Pro-level reasoning on a recurring basis, note that Pro access now requires Google AI Pro at $19.99.
- Set the effort level to “low” for routine, quick questions and save “high” for tasks that truly need deeper reasoning; Ground News reported in October 2026 that higher effort settings drain usage limits faster, so defaulting to “high” out of habit shortens how long a quota lasts.
- Gemini usage tied to business, school, or organizational Google accounts is reportedly governed by separate tiering, so check with an account administrator before assuming the personal-account restriction applies.
- Flash-Lite remains capable for short, straightforward tasks; the practical change mainly hits longer or more complex requests that benefited from Flash or Pro’s deeper reasoning.
Predictions: Where Gemini’s Free-Tier Access Goes From Here
Based on the pattern across 2026, from the May compute-accounting overhaul to the October model lockout, a handful of likely next moves stand out:
- Expect the new low/medium/high effort-level system to eventually carry its own tier restrictions, following the same path the model picker just took.
- Google AI Plus is likely to keep narrowing relative to AI Pro, since stripping Pro from Plus only makes sense as a lever if the company plans to keep separating those two tiers further over time.
- Rivals including OpenAI and Anthropic are unlikely to mirror a hard model lockout immediately, but compute costs affect every lab equally, so quieter tightening of free-tier caps elsewhere in the industry should not be ruled out over the next two quarters.
- Enterprise and education Gemini deployments will likely remain insulated from these consumer-tier changes for the near term, since Google has strong incentive to keep institutional customers on a predictable, separately negotiated footing.
- Expect continued friction and user complaints through the rest of the year as people discover the model picker has changed without necessarily having read Google’s support documentation in advance.
The Bigger Picture for Free AI Access
Google is not alone in facing this pressure, but it is the first of the major labs to respond with a model-level eligibility wall rather than a softer quota adjustment. That choice says something about where Gemini sits in Google’s broader AI strategy heading into 2027: less about maximizing free-tier reach for its own sake, more about using the free tier as a funnel into AI Plus, AI Pro, and AI Ultra. Whether that funnel converts well enough to justify the backlash from existing free users is the open question reporters and analysts will be watching now that the October 9 change has actually landed.
Frequently Asked Questions
When did Google’s free Gemini access change take effect?
October 9, 2026, according to reporting from 9to5Google and NokiaPowerUser that cited updated Google support documentation.
Which Gemini model can free users still use?
Gemini 3.5 Flash-Lite. Free accounts lost selectable access to Gemini 3.6 Flash and Gemini 3.1 Pro once the change rolled out on October 9.
Does Google AI Plus still include Gemini Pro after October 9?
Yes, for now. Google AI Plus, priced at $4.99 a month, keeps selectable Pro models until Google removes them from the plan on November 13, 2026. Google announced the change on October 16, and Flash and Flash-Lite access stays.
Which Google AI plan keeps full access to every Gemini model?
Google AI Pro at $19.99 a month and Google AI Ultra at $99.99 a month both retain Flash-Lite, Flash, and Pro, a lineup the Emirates News Agency (WAM) confirmed October 5. Pro also gained the Deep Think feature in this update.
Is this the same as the usage limits Google introduced earlier in 2026?
No. In May 2026, Google switched Gemini’s usage accounting from simple prompt counts to compute-based limits with a five-hour reset window and a weekly cap. That system reportedly remains in place. The October 9 change is a separate restriction on which models each tier can even select.
Does this affect Gemini on business, school, or enterprise accounts?
The restrictions described in current reporting apply to personal Gemini accounts. Organizational and enterprise plans are reportedly tiered separately, so check with an account administrator for specifics.
How does Gemini’s new free tier compare to ChatGPT’s free tier?
ChatGPT’s free tier gives limited access to more capable models and tools, with caps that vary by demand and feature rather than a hard lockout to a single model. Gemini’s October 9 change is a stricter, model-level restriction rather than a quota adjustment.
What are the new “effort level” settings in Gemini, and do they affect usage limits?
Reports describe new selectable “low,” “medium,” and “high” effort levels inside Gemini, which control how much reasoning effort a model applies to a response. Ground News reported in October 2026 that higher effort levels consume usage limits faster, so leaving a model set to “high” for simple questions burns through a quota quicker than using “low.” Exact per-level quotas still have not been published.
