Starting October 9, 2026, the Gemini free tier is limited to a single model: Flash-Lite. According to 9to5Google's report on October 3, people using the Gemini app without a Google AI subscription lose access to Flash and Pro, AI Plus subscribers lose Pro, and AI Pro subscribers gain Deep Think, which was previously reserved for Ultra. This piece lays out the tier-by-tier change, explains the economics of free-tier inference using Google's public price list, and compares the move with how Anthropic and OpenAI treat their own free users.
GeminiWhat changes for the Gemini free tier on October 9
Google's own model access and limits page states that starting in October 2026 there will be changes to model availability in personal accounts, and that the changes start for users without an AI subscription on October 9. The same page lists the end state: no subscription gets Flash-Lite only, AI Plus gets Flash-Lite and Flash, and AI Pro and AI Ultra get all three models. Deep Think, described as "maximum parallel reasoning," is listed for AI Pro and AI Ultra.
AI Plus does not switch on a single date. Per T-Break, Plus subscribers lose Pro on an account-specific date later in October, and 9to5Google reports that they will receive an email explaining when the change takes effect. Google also says that low, medium and high effort settings are coming to every available model in October. Higher effort completes harder tasks but uses more of a person's limit.
Google's published materials do not give a reason for the change, and they do not name a Flash-Lite version for the app, so the analysis below is inference from public pricing rather than a statement of Google's intent.
Gemini tiers before and after, side by side
The "before" column reflects what the reporting says each tier loses or gains; Google has not published a before-and-after table of its own.
| Tier | Price (US) | Before | After October 9 | Usage multiplier |
|---|---|---|---|---|
| Free | $0 | Flash-Lite, Flash, Pro | Flash-Lite only | Base allowance |
| AI Plus | $4.99 per month | Flash-Lite, Flash, Pro | Flash-Lite, Flash (Pro removed on an emailed date) | 2x |
| AI Pro | $19.99 per month | Flash-Lite, Flash, Pro | Flash-Lite, Flash, Pro, plus Deep Think | 4x |
| AI Ultra | $99.99 and $199.99 per month | Flash-Lite, Flash, Pro, Deep Think | Unchanged | 5x or 20x |
Prices come from T-Break and FourWeekMBA's summary of 9to5Google, and the multipliers come from Google's support page. Note that AI Pro is now the cheapest plan that includes Pro, which AI Weekly frames as a clearer ladder of tiers.
The economics of free-tier inference
Google moved the Gemini app to compute-based limits on May 17, 2026, according to its support page. Allowances refresh every five hours until a weekly limit is reached, and heavier prompts, features and longer chats consume more. The October change tightens the same lever from the other side: instead of only limiting how much a free user can do, it limits which model they can do it with.
Google's API price list shows how wide the gap between models is. Gemini 3.5 Flash-Lite costs $2.50 per million output tokens, Gemini 3.6 Flash $3.75 at its introductory rate, and Gemini 3.1 Pro Preview $12. API prices are not the app's internal cost, and the app does not publish which versions it runs, so this is a proxy for relative expense only.
The model the free tier keeps is the cheapest one on the price list
API list price per 1M output tokens, in US dollars. Flash-Lite sits at the bottom of the ladder, and the models removed from free and Plus accounts cost 1.5x to 4.8x as much per output token.
Gemini app access after October 9 shown in the labels above: Flash-Lite everywhere, Flash from Plus up, Pro from AI Pro up. Gemini 3.1 Pro is listed as a preview model; Flash introductory pricing runs through December 31, 2026.
By this proxy, each Pro response costs about 4.8 times as much as a Flash-Lite response of the same length, and Flash about 1.5 times. For a product serving a free audience, every conversation moved from Pro to Flash-Lite cuts the per-token bill sharply. Gagadget reads the change the same way, as cost control that also makes paid plans look more attractive.
Where Gemini 4 Argon fits
Gemini 4 Argon also bears on the picture. Google announced it on September 30 at an introductory $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 afterward. It is rolling out first to trusted cyber defenders, with paid API customers and AI Ultra subscribers to follow, as covered in our Gemini 4 Argon post. 9to5Google notes it is not clear where Argon will sit among Flash, Pro and the tiers above them.
A more expensive flagship at the top raises the value of the capacity underneath it. Every free-tier conversation that runs on Pro competes for the same accelerators as paid and API traffic. Whether that is the motive is not stated anywhere by Google, so it remains an open reading rather than a finding.
How this compares with Claude and ChatGPT
Rivals also hold their best models back from free users, though the shape differs.
- Claude. Anthropic's pricing page lists Sonnet and Haiku on the free plan and says free users cannot access Fable or Opus models. Pro costs $20 a month, or $17 with annual billing, and offers at least 5x more usage per five-hour session than Free.
- ChatGPT. According to Layer3 Labs, citing OpenAI's GPT-6 Astra announcement, Free and Go accounts have no GPT-6 Astra access, and Plus reaches Astra in Work and Codex rather than in chat.
The common pattern is that the free tier keeps a capable but cheaper model while the newest flagship stays behind payment. Gemini's difference is that the line moved down a rung for an existing audience, so it reads as a reduction for existing users.
What users can do
- Check the plan. Free accounts change October 9. AI Plus accounts get an email with their own date, so watch for it.
- Test Flash-Lite on real work. Short questions, rewriting and summaries are what it is designed for. Long documents and multi-step coding are where a smaller model is most likely to feel different.
- Weigh AI Pro against AI Plus. The Pro tier is the cheapest way to keep Pro, and it adds Deep Think and a 4x allowance.
- Use effort settings sparingly. Once they arrive, higher effort uses more of the limit.
- Keep a second option. Teams that depend on a stronger model can use a multi-model workspace such as Metir, so a change in one vendor's free tier does not set the model they work with.
Sources:
- 9to5Google: Gemini app limiting what models free and AI Plus users can access, AI Pro adding Deep Think
- Google Gemini Apps Help: model access and usage limits
- T-Break: Gemini model limits, free users lose Flash and Pro on 9 October
- AI Weekly: Google caps free Gemini at Flash-Lite starting October 9
- Android Headlines: Free Gemini accounts are losing Flash model access
- Gagadget: Google Gemini's free tier drops to Flash-Lite on October 9
- FourWeekMBA: Google limits free Gemini app to Flash-Lite from Oct 9
- Google AI for Developers: Gemini API pricing
- Google: Gemini 4 Argon
- Anthropic: Claude pricing
- Layer3 Labs: Is GPT-6 Astra free?
Image credits
- Hero: "Google Data Center, Council Bluffs Iowa" by Chad Davis, licensed CC BY 2.0, via Wikimedia Commons. It shows a Google data center in Iowa at sunset and is a file photo, not an image of any specific Gemini deployment.
- In-body: "Google Campus2 cropped" by Sebastian Bergmann, licensed CC BY-SA 2.0, via Wikimedia Commons.