Gemini Free Tier Cut to Flash-Lite on October 9

Google's help center updated a page this week with a quiet change: starting October 9, free Gemini users lose access to Flash and Pro, keeping only Flash-Lite. The same document covers the paid tiers, and the details matter more than the headline.

Short answer: Starting October 9, Gemini app users without a Google AI plan are limited to Flash-Lite. AI Plus subscribers ($4.99/month) keep Flash-Lite and Flash but lose Pro, with the timing arriving by email. AI Pro and AI Ultra keep the full set; AI Pro also gains Deep Think. Google did not touch the Gemini API, so your own key still reaches every model. On our fixed chat profile, the models the free tier loses cost $11.34 (Flash) and about $40 (Pro, 3.1) per 1,000 text messages on the API; Flash-Lite costs $6.54.

This is the second chapter of a change Google started in May, when Gemini Apps moved to compute-based usage limits that refresh every five hours. That first change re-priced attention. This one re-prices models: which ones exist at which tier, decided by billing code. Coverage so far has reported the change as a downgrade. It is one, but the interesting part is what it costs to route around it, and Google left the cheap door open.

What changes on October 9 for the Gemini free tier

The support page says the changes start taking effect for users without an AI subscription on October 9. For AI Plus subscribers, the page promises "an email that explains when these changes will take effect" — so that downgrade has no public date yet.

Model availability after the changes, per the page and 9to5Google's read of it:

Google AI plan	Flash-Lite	Flash	Pro
Without a plan	✓	—	—
AI Plus ($4.99/mo)	✓	✓	—
AI Pro ($19.99/mo)	✓	✓	✓
AI Ultra ($99.99+/mo)	✓	✓	✓

Free users currently get 3.6 Flash and varying amounts of 3.1 Pro; after October 9 the free tier is Flash-Lite only, which 9to5Google identifies as the 3.5 generation. Model versions matter here: the free tier goes from three models down to one, and the model that remains is a generation or two older. AI Pro and Ultra see no regressions, and AI Pro gains the Deep Think option previously reserved for the $99.99-and-up plans.

Two more details from the same pages. Every available model gets new low, medium, and high effort levels that use more of your limit as they climb. And the model everyone keeps asking about, Gemini 4 Argon, goes to AI Ultra subscribers first, so it won't rescue any of these tiers when it ships.

What the cut removes, priced

We priced every model in the table on the same fixed profile we use across this blog: 8,000 input tokens, 2,000 output tokens, 40% of the input hitting the prompt cache. That is one normal message in a long conversation.

Model	$/1M in	$/1M out	1,000 messages
Flash-Lite (3.5)	$0.30	$2.50	$6.54
Flash (3.6 / 3.8)	$0.75	$3.75	$11.34
Flash from 2027	$1.50	$7.50	$22.68
Pro (3.1)	$2.00	$12.00	~$40

Rates read from ai.google.dev on October 4, 2026. The Flash row is promotional through December 31, 2026; the model doubles on January 1, which we walked through in September. The Pro row uses the $2/$12 card Google cross-references in its own pricing tables (the model's dedicated section is gone from the page), and it has no published cache rate anymore, so we billed every input token at full price here. The build for the Flash row: 1,000 messages × 8,000 input tokens = 8M input, 40% of it cached at $0.075 and the rest at $0.75, plus 2M output at $3.75. Total: $11.34.

So the practical size of the free-tier cut, in API dollars, is $4.80 per 1,000 messages (the Flash-to-Flash-Lite gap). The most capable model leaving the free roster, 3.1 Pro, costs roughly six times what Flash-Lite does per message.

What nobody tells you about the new Gemini tiers

The API's free tier is untouched. Every outlet covering this story points at AI Pro's $19.99 as the fix. But the Gemini Developer API is a separate product for this change, and its free tier still lists free-of-charge access on Flash-class models, with paid rates starting at $0.30/$2.50 per million tokens. The catch has always been the same: API free-tier content is used to improve Google's products, and rate limits are tight. For a personal account that sends a few hundred messages a month, though, the API route replaces the cut models for single-digit dollars, no subscription.

"One model" now does three jobs. The new low/medium/high effort levels let you push Flash-Lite harder on the same account. That does not make it a Pro substitute; the tiers differ in capabilities and effort levels cannot close that gap. But the practical hit for casual users is smaller than the three-tier table suggests, and heavier for anyone who leans on the high setting.

The AI Plus change has no date. Free users have October 9 circled. Plus users have an email to wait for. If you are on AI Plus for Pro access specifically, budget for the downgrade now instead of reacting to the email.

The API math if you just want Flash and Pro back

Google's plan prices and pay-per-use rates meet somewhere, and the meeting point is easy to compute. $4.99 buys about 440 Flash messages per month on the API, or about 760 on Flash-Lite. $19.99 buys about 1,760 Flash messages, or roughly 3,060 Flash-Lite ones. If your Gemini usage is mostly text, those are the trade-off numbers: a flat plan for the full app experience, or per-message pricing that only charges what you send.

The honest counterweight: AI Pro bundles media generation, Deep Research, storage, and higher limits that a text-only API comparison ignores. If you use those, the subscription is the better deal and this section is not for you.

For the text-heavy cases, the BYOK path is the one this blog exists for. Connect a Google API key to a workspace like ours (the setup guide takes about three minutes) and every model in the table above is a dropdown entry, billed by Google at the rates listed. No plan tiers decide which model you get.

Which Should You Choose?

If you...	Best option	Why
Ask Gemini a few questions a day	Stay free	Flash-Lite covers light, everyday use
Want Flash on a near-zero budget	Gemini API free tier	Free-of-charge access on Flash-class models, low limits
Used Pro on the free tier for real work	BYOK, pay per use	~$40 per 1,000 Pro messages only when you send them
Want media generation and Deep Research	AI Pro ($19.99/mo)	Keeps every model, adds Deep Think, bundles the app
Switch models and providers in one place	Multimodel Chat + your keys	One workspace, providers bill you directly

What we couldn't verify

Plan prices ($4.99, $19.99, $99.99) come from Google's support pages as read by us and reported by 9to5Google and The Decoder; Google's plans page renders its dollar figures client-side, so we could not screenshot them directly.

The free tier's previous Pro access was already "varying" before this change, so we cannot put an exact number on what a given free user loses. The 3.1 Pro rate is reconstructed from cross-references on Google's pricing page; its cache rate is not published anymore. And one flagged guess: that this tier reshuffle quietly prepares the billing ground for Argon's cost structure. That is analysis, not a statement from Google, and we will re-check it when Argon ships.

Workspace accounts are a separate question that came up in the community threads we read; the support pages we checked cover personal accounts only.

FAQ

What changes for me on October 9? If you use Gemini with a personal account and no Google AI subscription, you keep Flash-Lite and lose Flash and Pro. If you pay for AI Plus, you lose Pro and will get an email with your timing.

Does this affect the Gemini API or my API key? No. The change is specific to the Gemini apps. The API's free tier, paid rates, and model list are unchanged as of October 4.

Is Flash-Lite good enough to just live with? For light daily use, close. It is Google's smallest, cheapest model, and the new effort levels stretch what it can do. For long documents, hard reasoning, or agentic work, you will notice the gap quickly; that was what Pro access bought, even in "varying amounts."

Should I upgrade to AI Pro for $19.99? Only if you use the bundle: media generation, Deep Research, and the higher limits. If your usage is text and you mostly want model choice back, the pay-per-use numbers above beat it. About 440 Flash messages fit inside $4.99.

What is the cheapest way to keep Flash and Pro? A Google API key. Flash bills at $0.75/$3.75 per million tokens (promo through December 31; $1.50/$7.50 after), Flash-Lite at $0.30/$2.50, Pro at roughly $2/$12. Pay for what you send, and keep app-subscription decisions separate from model-access decisions; the general crossover math between API and subscriptions is here.

---

Start your free trial → — 7 days, all providers, no credit card required.