Google has announced that the AI models available in the Gemini app for personal accounts will now vary by subscription plan. Starting October 9, 2026, free users will have access only to Flash-Lite. Google AI Plus subscribers will be limited to Flash-Lite and Flash from an effective date that will be notified to each user individually. Google AI Pro and AI Ultra subscribers will continue to be able to choose among Flash-Lite, Flash and Pro.
Google AI Pro will also gain access to Deep Think, which Google's consumer help pages have so far described as exclusive to Ultra. From now on, the plan you choose will matter not only for usage limits but also for which models and reasoning features you can use. Looking at monthly prices in Japan and the amount of context each plan can handle at once, the positioning of Plus and Pro also changes.
Free plan changes on October 9; Plus users will be notified of their date
After the change, free users will no longer be able to choose Flash or Pro. AI Plus will keep Flash but lose access to Pro. Only AI Pro and AI Ultra will be able to select all three models: Flash-Lite, Flash and Pro.
Google's October announcement says the change applies to the Gemini app when used with a personal account.
Because the names are similar, it helps to distinguish the "Google AI Pro" subscription plan from the "Pro" AI model. What was announced is not the retirement of the Pro model itself, but a change in which plans can select it. From this announcement alone, it is not possible to tell whether the same conditions apply to work or school accounts or to the Gemini API.
The change for free users begins on October 9. AI Plus subscribers will receive an email telling them when the change will apply to their account, so Plus will not switch uniformly on October 9.
When reading the official information, it is worth separating current conditions from post-change conditions. As of October 4, the consumer usage-limits help page still shows a table in which all plans can use Flash-Lite, Flash and Pro. Sales pages for Japan likewise still say the free plan has limited access to Pro.
The October change announcement, by contrast, spells out the new model-access conditions. Being able to choose a model in the Gemini app today does not mean you will still be able to use it after the change takes effect.
What differs between Plus at ¥725 per month and Pro at ¥2,900
The standard prices in Japan are ¥725 per month for Google AI Plus and ¥2,900 per month for Google AI Pro.
Combining the post-change model access and usage limits that Google has presented with the context window available at once, the plans compare as follows.
| Plan | Monthly price shown in Japan | Models available after the change | Usage limit | Context window |
|---|---|---|---|---|
| Free | ¥0 | Flash-Lite | Standard | 32,000 tokens |
| AI Plus | ¥725 | Flash-Lite, Flash | 2x standard | 128,000 tokens |
| AI Pro | ¥2,900 | Flash-Lite, Flash, Pro | 4x standard | 1,000,000 tokens |
| AI Ultra 5x | ¥14,500 | Flash-Lite, Flash, Pro | 5x AI Pro | 1,000,000 tokens |
| AI Ultra 20x | ¥32,000 | Flash-Lite, Flash, Pro | 20x AI Pro | 1,000,000 tokens |
Models and usage limits are based on Google's change announcement, prices on the Japan plan page, and context windows on the consumer help page. Prices and context windows reflect what was shown as of October 4, 2026, and the model column reflects conditions after the change takes effect.
The change announcement does not mention any change to context windows. Also, usage limits of "2x standard" and "4x standard" do not mean that the number of questions you can send or the quality of answers doubles or quadruples.
Under the free plan after the change, the only model available at the standard usage limit will be Flash-Lite. AI Plus doubles the usage limit relative to standard, but offers just two models, Flash-Lite and Flash. The size of the usage limit and the models you can use are separate conditions.
The most affected will be users who have so far used Pro on AI Plus. Moving from free to Plus lets you choose Flash, but Pro will be unavailable. To keep selecting Pro after the change, you will need AI Pro or higher.
Moving from Plus to Pro raises the monthly price by ¥2,175, making it four times as expensive. The usage limit goes from 2x to 4x standard, so it is twice that of Plus.
However, judging the price gap by usage alone overlooks the differences in access to the Pro model and in the context window, which expands from 128,000 tokens to 1,000,000. Conversely, if Flash handles your everyday work well, you do not necessarily need to move to a higher plan just to use the Pro model.
Usage allowances reset every five hours; model access does not
Gemini's usage limits are not determined simply by how many questions you send.
Since May 2026, Google has moved to usage limits based on compute. Consumption is affected by the complexity of the prompt, the model and features used, and the length of the chat. Media generation, Deep Research, the Pro model, extended thinking and Deep Think are listed as features that consume more of the allowance.
The regular usage allowance resets every five hours, but there is also a separate weekly limit. The conditions for regaining access differ depending on whether you have used up only the five-hour allowance or have reached the weekly cap. It is therefore not the case that waiting five hours will always restore every feature.
Models that free and AI Plus plans lose access to under this change will also not come back when the allowance resets. Usage limits and access rights to models are separate.
The October announcement also describes a setting that lets you choose an "effort level" of low, medium or high for the available models. Raising the effort level lets the model do more processing and produce more detailed answers, but it also consumes more of your allowance.
This setting adjusts how much processing the available model does; it is not a feature for changing the model itself. For example, setting a high effort level on Flash-Lite does not grant access to the Pro model.
For that reason, you also cannot work out how many questions you can send per day from the "2x standard" figure for AI Plus alone. Repeatedly summarizing short texts and feeding in a large volume of material for sustained complex reasoning can use different amounts of the allowance even with the same number of requests.
When choosing a plan, it is worth looking not just at the number of questions but at how much of the allowance the models and features you normally use consume.
Deep Think opens up to Pro, but differences with Ultra remain in usage
A major change for AI Pro is access to Deep Think.
The consumer usage-limits help page as of October 4 describes Deep Think as "AI Ultra only." The October announcement, however, says that AI Pro or AI Ultra subscribers can use the Deep Think option, which performs maximal parallel reasoning.
This means an advanced reasoning feature previously limited to Ultra will extend to the less expensive Pro plan.
The announcement does not state how many times AI Pro users can use Deep Think, or how much of the allowance a single run consumes. Having access to Deep Think and being able to use it enough on a regular basis should be considered separately.
AI Ultra has a usage limit of either 5x or 20x that of AI Pro, depending on the plan. Even with Deep Think opened to Pro, users will not be able to use the same volume as Ultra.
Whether Pro is enough for users who occasionally hand complex problems to Deep Think, or whether frequent users need Ultra's larger allowance, will come down to checking actual consumption.
If you're reconsidering your plan, start with the models you actually use
Google positions Flash-Lite as a fast, efficient model suited to everyday tasks such as summarizing and brainstorming. Flash balances speed and reasoning ability to handle a wide range of problems, while Pro is the top-tier model, strong at complex math, coding and similar tasks.
This change alone does not allow one to derive figures such as how much the quality of answers will fall on the free plan. What matters is checking which model your usual tasks actually require.
If your use is mainly short summaries and brainstorming, you can check whether Flash-Lite, which remains available after the change, is enough. AI Plus is an option for those who want to use Flash day to day. Users who have so far handed math, programming, complex analysis and similar work to the Pro model on Plus will need to decide, before their effective date, whether Flash can substitute or whether to move to AI Pro.
When handling long documents or large amounts of code, the difference in context windows also matters.
The context window indicates how much information, such as prompts, conversation and files, Gemini can refer to at once. AI Plus offers 128,000 tokens versus 1,000,000 for AI Pro, a large gap. However, being able to handle more information is not the same as getting more accurate answers.
The 400GB of cloud storage included with AI Plus and the 5TB included with AI Pro are also separate from the amount of information Gemini can read at once.
AI Plus subscribers should check the email from Google for their effective date, and free users should recheck which models they can use from October 9. Google also recommends updating the Gemini mobile app to the latest version.
After the change, checking whether you can select the models you normally use, handle the length of material you need, and have enough usage to finish your work will make it easier to judge the differences between plans that monthly prices alone do not reveal.