I asked Gemini a question that I wanted her to think deeply about, but for some reason I couldn't select the thinking mode. Or have you ever received a standard answer without any thought process displayed?
As of February 2026, Google's AI chat "Gemini" has three modes: Fast mode, Thinking mode, and Pro mode. Among them, the selling point of Thinking Mode is that ``AI makes a careful inference before answering,'' but there are many cases where it cannot be used or displayed due to the limit on the number of times it can be used or problems with the settings.
In this article, we will organize the reasons why Gemini's Thinking Mode cannot be used into 5 reasons, and we will also explain tips on how to make the most of Gemini's Thinking Mode even for free users.
What exactly is Gemini's "thinking mode"?
Gemini's thinking mode is a function that allows you to assemble your own "logical reasoning steps" before the AI provides an answer. Roughly speaking, the AI automatically does the work that humans do when solving difficult problems, such as writing intermediate formulas on a notepad.
As of February 2026, the Gemini app has the following three modes.
| Mode | Internal model | Characteristics | Suitable use |
|---|---|---|---|
| Fast | Gemini 3 Flash | Fastest. For light questions | News summaries, daily research |
| Thinking | Gemini 3 Flash Thinking | Strengthen the reasoning step | Contract interpretation, logical judgment |
| Pro | Gemini 3 Pro | Highest performance. Multimodal support | Code analysis, judgment including images and videos |
The point is that Thinking mode and Pro mode are different. Thinking mode specializes in ``depth of reasoning,'' and Pro mode has ``high overall processing ability.'' If you want intelligent answers, choose Pro, and if you want people to think carefully and logically, choose Thinking Mode.
5 reasons why thinking mode cannot be used or displayed
Cause 1: The free version limit has been reached
This is the most common cause. As of February 2026, according to Google's official help, the upper limit for the use of thinking mode for each plan is as follows.
| Plan | Monthly fee | Thinking mode (per day) | Pro mode (per day) |
|---|---|---|---|
| Free | 0 yen | 3 to 5 times | Very limited |
| Google AI Plus | Around 1,350 yen per month | About 90 times | About 30 times |
| Google AI Pro | Around 2,900 yen per month | Approx. 300 times | Approx. 100 times |
| Google AI Ultra | Around 5,900 yen per month | Approx. 1,500 times | Approx. 500 times |
The free version can be used only 3 to 5 times a day. Moreover, during peak times (night to late night Japan time) when access is concentrated, it may not be displayed as an option in the first place.
Workaround: The only option is to wait for the number of times limit to be reset. When the upper limit is reached, it will automatically switch to Fast mode, so you can select Thinking mode again the next day.
Cause 2: Thinking is not selected in model selection
When you open the Gemini app, there is a model selection dropdown at the top of the chat input field. If "Fast" is left selected here, the thought process will not be executed.
What to do: Tap (click) the Model Switch button at the top of the chat screen and select "Think". In the case of the smartphone app, the model name is displayed small and is easy to miss, so be careful.
Cause 3: Not aware of "folding" of thought process
When you answer in thinking mode, a collapsible section called 'Thought Process' will appear at the top of your answer. Tap this to see what kind of inference the AI made.
However, since this folded section is closed by default, there are many cases where people mistakenly think that the thought process is not displayed!
What to do: When you receive an answer, try tapping the "Thought Process" label above the text. The AI reasoning process should unfold.
Cause 4: The browser cache/app version is outdated
If you are using the web version of Gemini (gemini.google.com), the UI may not display correctly due to browser cache. Also, if the smartphone app version is old, the new model switching UI may not be reflected.
How to deal with it:
- Web version: Clear your browser cache and reload the page
- Smartphone version: Update the app to the latest version on App Store / Google Play
- If neither works, try accessing gemini.google.com in incognito mode (private browsing)
Cause 5: Regional/language settings problem
The availability of some Gemini functions varies depending on region and language settings. As of February 2026, Thinking Mode itself is available in Japanese, but if your Google account's regional settings are set to a country other than Japan, some functions may be restricted.
What to do: Check if the region is set to "Japan" in Google account settings.
3 tips to make the most of thinking mode even in the free version
To be honest, the free version's limit of 3 to 5 times a day is tough. However, if you use it wisely, you can get good results even with a small number of uses.
Tips 1: Increase the accuracy of one question with "structured prompts"
The trick to questions that get you into thinking mode is to write a set of background information, goals, and constraints. If the question is ambiguous, the AI's reasoning will be off the mark, and a valuable session will be wasted.
Bad example: “Is this contract okay?”
Good example: "For the following business outsourcing contract, please list three clauses that are disadvantageous from a freelance contractor's perspective and suggest amendments for each. I would like you to pay particular attention to the terms of compensation for damages, intellectual property rights, and contract termination."
Tips 2: Fast mode for everyday questions, thinking mode for important decisions
It would be a waste to use thinking mode for simple questions like "What's the weather like today?" or "What does this English word mean?" Basically, default to Fast mode and switch to Thinking mode only in the following situations.
- Confirmation of contract/terms of use
- When you want to make a decision by comparing multiple options
- When you want to analyze the cause of a program bug
- When you want to check the logical structure of a report or paper
Tips 3: Utilize “Thinking Levels” added in January 2026
In the January 2026 update, a Thinking Level adjustment function has been added to Thinking Mode. There are two levels, "Low" and "High", which can be used depending on the difficulty of the task.
- Low: For simple reasoning tasks. Consumes less “thinking quota”
- High: For complex analysis and multi-step reasoning. High quota consumption but high accuracy
If you use High for all tasks, you will reach the limit quickly. It is wise to try Low first, and if the results are not satisfactory, try again with High.
Should I upgrade to a paid plan? Judgment guideline
If you have situations where you want AI to think deeply more than 10 times a day, it is worth considering upgrading to a paid plan (Google AI Plus or higher).
The limitations of the free version tend to become a bottleneck, especially for the following people.
- I use AI to perform code reviews and document checks on a daily basis at work
- I want to ask AI to check contracts and invoices for my side job or tax return
- I want to frequently check the logic of university reports and papers
On the other hand, if you only occasionally ask difficult questions, 3 to 5 times using the free version is sufficient. First, try using the free version for about a week, and if you feel like you want to use it more but are stopped by the restrictions, consider upgrading.
As of February 2026, Google AI Plus costs approximately 1,350 yen per month, and Google AI Pro costs approximately 2,900 yen per month. For more information, please check the latest pricing on Google AI subscription page.
FAQ
What is the difference between Gemini's thinking mode and fast mode?
Fast mode uses Gemini 3 Flash and answers with an emphasis on speed. Thinking mode uses Gemini 3 Flash Thinking and takes logical reasoning steps before answering, increasing accuracy in complex questions and situations that require judgment. However, it may take a few seconds to more than 10 seconds to respond.
When will the thinking mode limit be reset in the free version?
As of February 2026, the limit is reset approximately every 24 hours. However, Google's official help does not announce a clear reset time, so if you find that it worked yesterday but still doesn't work today, please wait a few hours and try again.
Will Google learn what I hear in Thinking Mode?
According to Google's Privacy Policy for Generative AI, input data will not be used for model training in Google Workspace paid plans. However, in the free version of the Gemini app, your input may be used for quality improvement. If you handle confidential information, please consider using a paid plan.
Can I use thinking mode via API?
Yes. Gemini API allows you to control the thinking function with the thinkingConfig parameter. Detailed implementation methods are described in Google AI for Developers official documentation. If using the API, the fee structure (token charge) will be different from the app version.
References
- Usage limits and upgrades for Gemini apps depending on your Google AI subscription plan — Google Gemini Help
- Gemini thinking | Gemini API — Google AI for Developers
- Google Workspace Generated AI Privacy Hub — Google Workspace Admin Help
- Google AI Pro & Ultra — get access to Gemini 3.1 Pro & more — Google
