Gemini App Sets Model Availability by Plan from October 9 - Flash-Lite Is the Only Pick Without a Subscription
Google's help page says model availability in the Gemini app will be set by plan. The change starts taking effect for users without a subscription on October 9, after which Flash-Lite is the only pick. AI Plus has Flash-Lite and Flash, with its start date sent by individual email.
Google’s help page states that the models a personal account can pick in the Gemini app will be divided up by plan1. For people who are not on an AI subscription, the change begins on October 9, after which Flash-Lite is the only option left.
After the change, AI Plus has two models to pick from, Flash-Lite and Flash; Pro is not among them. AI Pro and AI Ultra have all three. The timing differs by plan: for AI Plus, the page only says that an email will explain when the change takes effect1.
Which models each plan keeps
The table on the help page describes the state after the change as follows1.
| Google AI plan | Flash-Lite | Flash | Pro |
|---|---|---|---|
| Without a plan | yes | — | — |
| AI Plus | yes | yes | — |
| AI Pro | yes | yes | yes |
| AI Ultra | yes | yes | yes |
The three column headings are “Flash-Lite”, “Flash” and “Pro”, with no version numbers attached. Which release of Flash or Pro is meant therefore cannot be read off this notice.
The page also says an effort level of low, medium or high can be chosen for each available model. Google explains that this raises the model’s ability to finish tasks and produces more thorough answers, while also consuming more of the limit. Deep Think, for maximum parallel reasoning, is described as available to AI Pro and AI Ultra subscribers. On the API side, a configuration offering a reasoning-heavy variant at the same rate had already appeared with Gemini 3.8 Live Extended Thinking.
Limits are counted in compute
The same page covers how usage limits are counted, in text placed under a section headed “Previous changes.” Gemini moves to compute-based limits that refresh every five hours until the weekly limit is reached, and the calculation factors in prompt complexity, the features used, and chat length1.
What the page flags as burning through the limit sooner is generating media — pictures, video, music — plus Deep Research, extended thinking, Deep Think and the Pro model.
Per-plan limits are expressed as multipliers. Without a plan is “standard limits”, AI Plus doubles that, AI Pro quadruples it, and AI Ultra sits at either five or twenty times the AI Pro figure, according to which subscription is held1. The underlying standard limit is never given as an absolute figure, so the multipliers cannot be worked back to a concrete number.
Limits stated as multipliers have been hard to read before now. When Anthropic changed Claude Code’s weekly limits in September, a permanent 25% raise against the original allowance landed about 17% below the level in force once the temporary increase ending that same day was taken into account. Depending on where the comparison is anchored, the same change can be written as an increase or a decrease. The help page gives no formula for compute, and no conversion into a published unit such as a token.
Not the first revision
The help page carries a section headed “Previous changes.” It records an earlier revision of usage limits and of which models were offered, dated May 17, 2026 for users over 18, with those under 18 moved onto it on July 241. That makes this at least the second revision.
Over that period, the app’s own numbers have grown. In August, Google announced that the Gemini app had passed one billion monthly users. The models have kept turning over as well, and Gemini 4 Argon, announced at the end of September, was to begin its broad rollout with paid API customers and Google AI Ultra subscribers. Handing access to higher plans first appears to line up with the way models are being divided here.
What subscribers still cannot tell
On Google’s plan page, Japanese pricing as of October 5, 2026 shows AI Plus at 725 yen per month, AI Pro at 2,900 yen, and AI Ultra from 14,500 yen. The comparison table on that page splits AI Ultra into two steps, AI Ultra 5x at 14,500 yen and AI Ultra 20x at 32,000 yen2. The decision is whether to stay on the free tier, move up to AI Plus at 725 yen, or go as far as AI Pro at 2,900 yen.
For anyone weighing AI Plus, though, the information is not all there. The help page table does not include Pro for AI Plus after the change, yet the AI Plus description on the plan page, as of the same October 5, still carries a line about expanded access to Gemini 3.1 Pro and Gemini’s Deep Research, and the comparison table on that page marks AI Plus as “added” on its row for access to Google’s Pro models2. Neither page explains how those two descriptions correspond.
On Deep Think the two pages flatly disagree. The help page names AI Pro and AI Ultra subscribers as having access1. The comparison table’s Deep Think row, meanwhile, marks AI Plus and AI Pro as unavailable and AI Ultra 5x and AI Ultra 20x as available2. Neither page says which one reflects the current state.
The timing is also unstated for AI Plus. Because notice arrives as an individual email, subscribers will not know when their own environment changes until it reaches them.
For free use, at least, the date is firm. The change starts taking effect for users without a subscription on October 9, after which Flash-Lite is the single model available. Anyone leaning on the Gemini app for work that Flash-Lite cannot cover has to decide between paying and moving elsewhere.
Google’s help page carries no publication or update date, and offers no indication of timing beyond the phrases about October 2026 and October 9 in its own text. The Japanese version of the page also carries a note that it may contain AI-translated content.
Sources
- Changes to Gemini model access and limits - Google official help (Gemini Apps)
- Google AI plans - Google’s official plan page (Japanese pricing as shown on October 5, 2026)
Was this article helpful?
Thank you!
Received. Thank you!