Gemini's Free Tier Becomes a One-Model Product on October 9

Starting October 9, 2026, the free tier of the Gemini app has exactly one model left: Flash-Lite. Gemini Flash (3.6) and Gemini Pro (3.1) disappear from free accounts the same day, and Google's $4.99/month AI Plus plan loses Pro access too — full access to all three models now starts at AI Pro. This page is a plain record of what changes, what the app's limits have already done over the past six months, and why Google is doing this now.

What exactly changes on October 9

Per Google's own help documentation and the coverage that followed it (9to5Google on October 3, Engadget on October 6, plus TechRadar and Mashable earlier in the window), the model matrix for personal accounts becomes:

Plan	Price	Models from Oct 9
Free	$0	Flash-Lite only
Google AI Plus	$4.99/mo	Flash + Flash-Lite (Pro removed)
Google AI Pro	paid tier	All three
Google AI Ultra	paid tier	All three, higher limits

Two things worth stating precisely. First, this is a model-access cut, not a usage-limit cut — a free user who was happy with Flash-Lite loses nothing on October 9. Second, the AI Plus change is the sharper one for Google's relationship with its own paying users: that tier keeps its $4.99 price while its model bundle shrinks. Google is not quietly repricing the cheap tier; it is quietly devaluing it.

Confidence: high. The October 9 date and the per-plan model list are corroborated across four independent outlets plus Google's own updated help pages, which had already been edited to reflect the new matrix by October 6.

The six months before it

October 9 is not an isolated event. The verifiable sequence for Gemini Apps this year, with the sources for each step:

• May 17, 2026 — Gemini Apps switched to compute-based usage limits, replacing fixed daily caps like "N prompts per day." Google's own help page — "Gemini Apps limits and upgrades for Google AI subscribers" — documents the date directly. The significance: once usage is measured in compute rather than in prompts, the mix of models you touch silently changes how fast your allowance burns. A session on Pro weighs more than the same session on Flash-Lite — which is exactly the lever October 9 formalizes.
• September 2026 — community reports of free-tier and even paid-tier frustration with limits sharpened. A support thread dated September 10 shows a Pro subscriber hitting a "reached your limit on 2.5 Pro" wall and being told to upgrade for higher limits. Confidence: medium — this is community evidence of pressure, not an official policy change, so treat it as color, not data.
• October 3–6, 2026 — tech press begins reporting the October 9 model-access change from Google's announcement; by October 6 Google's own plan pages had been updated to the new matrix. Confidence: high.

Read in order, the sequence is coherent: first make usage metered and opaque (May), then make the expensive models unreachable below the paid tiers (October). Whether that ordering was deliberate or opportunistic, only Google knows. The direction, however, is documented twice over.

Why now — the boring answer is probably the real one

The tempting narrative is that Google is squeezing users because it can. The boring answer fits the evidence better: Pro-class reasoning models are genuinely expensive to serve, free-tier usage of them is the most subsidized thing the Gemini app does, and May's shift to compute-based metering gave Google the telemetry to see exactly how much of that subsidy was going to whom. Once you can measure it, cutting it is a spreadsheet decision.

It also lines up with what the same company did on the API side. Our Gemini 3.8 Flash API pricing breakdown covers the parallel move there: the introductory $0.75/$3.75 rate doubles on January 1, 2027, and thinking tokens bill as output. The app tiers and the API are converging on the same posture — the cheap-access window was a land-grab tool, and the land grab is over.

One caution against over-reading: none of Google's pages state a motive, and no executive has gone on record with one. The cost-pressure explanation is inference from a documented pattern, not a documented fact.

If you were riding the free tier

Three honest options, cheapest first:

• Stay and adapt. If Flash-Lite covers your actual use — most quick lookups and drafts — nothing changes for you on October 9. The cut only bites if you were using free Pro for reasoning-heavy work.
• Move that work to the API's free tier. The Gemini API still has a free tier with published rate limits, on different terms from the app. It's rate-limited and not a drop-in replacement for the app experience, but for scripted or batch-shaped tasks it remains free access to capable models. What Google does to the app tier does not automatically extend there — yet.
• Pay, but pick deliberately. If you upgrade, note what October 9 actually did: AI Plus ($4.99) no longer buys Pro. The cheapest plan that includes it is AI Pro. If you were paying $4.99 mainly for model access, check whether the tier still justifies itself for you — and if your heavy use is API-shaped, compare subscription plans against metered API spend before committing. For raw per-token economics across vendors, our cheapest LLM API provider comparison is the reference page.

The question this leaves open

The number to watch is AI Plus retention. Google cut the model bundle of a paid tier without cutting its price, betting that $4.99 subscribers mostly want storage (the tier includes 400 GB) and convenience, not Pro-class reasoning. If churn on that tier spikes in the following weeks, Google will have learned that the people paying $4.99 were paying for the model — and the next move could go either direction: a price cut, or a further narrowing. If retention holds, expect the same scissors to come back for whatever slack remains at the bottom of the ladder.

What is not open to interpretation is the date. October 9 is the day the free tier becomes a one-model product.

Sources: 9to5Google — Gemini app limiting what models free & AI Plus users can use (Oct 3, 2026) · Google One — AI plans (updated matrix) · Google support help page "Gemini Apps limits and upgrades for Google AI subscribers" (May 17 compute-limits date; linked from Gemini app settings)