Google narrows free Gemini to a single model: from October 9, only Flash-Lite

How the free tier changes. The official page is about model access and usage, and Google's usage table spells out every tier: users without a plan sit at standard limits, AI Plus gets 2x standard limits, AI Pro 4x, and AI Ultra 5x or 20x AI Pro depending on the subscription. According to that page and Google's official Gemini subscriptions page, accounts without a subscription are narrowed to Flash-Lite; per tech coverage of Google's notice and subscriptions page, AI Plus keeps Flash but loses Pro with the effective date communicated by email, AI Pro and AI Ultra keep all three of Flash-Lite, Flash and Pro, and US monthly pricing is $4.99 for AI Plus and $19.99 for AI Pro. In other words, keeping Pro as an option now starts at AI Pro rather than at nothing.

The gap between the three models decides how much this actually matters. Google's own description of model roles: Flash-Lite for fast everyday work such as summarizing material and brainstorming, Flash for a balance of speed and reasoning, and Pro for complex math, coding and deep understanding of files, images and video, at the cost of slower responses. These are task tiers, not marketing tiers. A free user who used to pick Flash or Pro for hard questions now either accepts Flash-Lite or pays, which is exactly why the change drew attention.

Easier to miss than the model list is the allowance math. The official page states Gemini moves to compute-based usage limits that refresh every 5 hours until the weekly limit, and that the calculation factors in prompt complexity, features used and chat length. The unit of the limit is therefore not how many questions you asked but how much compute you burned. Google also notes premium models and features consume more usage and can push you to your limit faster, which explains a complaint people often have: the same number of prompts can drain the budget at very different speeds.

The official usage table ties payment to ceilings: standard limits without a plan, 2x standard for AI Plus, 4x for AI Pro, and 5x or 20x AI Pro for AI Ultra depending on the subscription. That table explains two things. First, Google is selling a ceiling as well as model tiers. Second, access and allowance are two sides of one rule: you can ration Flash-Lite, or you can pay for a higher ceiling.

The parallel-reasoning option Deep Think is still listed as AI Ultra only. 9to5Google reported on October 3 that AI Pro would gain it, but Google's October change note and general limits page both listed it under AI Ultra as of October 6, with the latter explicitly calling it Ultra-only, so the two accounts do not line up. Looking back, Google introduced Gemini 3.6 Flash and 3.5 Flash-Lite in July 2026 and offered students a free year in August: eligible US college students get AI Pro, and eligible students in some non-US markets get AI Plus, with those accounts holding a subscription even during the no-charge period. Taken together, the direction is to pull stronger models into paid tiers while using student offers to hold penetration.

For everyday use, the practical impact of this change is limited. Google's page is explicit about what each model is for, and the one that survives, Flash-Lite, is named Lite for a reason: it covers summarizing, drafting and brainstorming. What actually breaks is the pattern of a free account occasionally reaching for Pro on complex math or code. The other detail worth noting is that the unit of the limit becomes compute: Google warns prompt complexity, features used and chat length all count toward consumption, so asking a complete question once and avoiding very long threads saves more than trial and error. Together those two points are the most practical adjustments an ordinary user can make.