Key Takeaways
- Which AI Model Are You? is a free browser quiz from developer Lattice that labels you as an AI model such as GPT, Claude, GLM or Qwen, with a short written reason attached.
- It runs at ai.tatoe.top on phones and desktop browsers, requires no login or personal details according to the available information, and takes only a few minutes to finish.
- Results are playful labels rather than measurements: a GLM-5.3 verdict says nothing about your abilities and nothing about that model's real performance.
- The fun is in comparing persona lines with friends, so share the screenshot, argue about the labels and treat the outcome as entertainment rather than a scientific assessment.
Quick Facts
| Type | Free browser personality quiz |
|---|---|
| Devices | Phones and desktop browsers |
| Price | Free, no login required |
| Languages | Primarily Japanese; check the official site |
| Highlight | Flattering, dramatic persona lines |
| Learning curve | None; a few minutes to finish |
| Category | Entertainment |
| Creator | Lattice |
| Official site | ai.tatoe.top ↗ |
| Heat | 🔥 1K Heat |
| Last updated |
A Free Browser Quiz That Labels You as an AI Model
Which AI Model Are You? is a free browser quiz from the developer Lattice that answers a very 2026 question: if you were an AI model, which one would you be? Instead of sorting you into Hogwarts houses or Myers-Briggs letters, it hands you a model name such as GPT, Claude, GLM or Qwen, plus a short written reason for the verdict. It arrived on October 18, 2026, and quickly became a talking point among Japanese-speaking AI enthusiasts who enjoy comparing labels with friends.
The whole thing lives at ai.tatoe.top and runs in an ordinary web browser. There is nothing to install, no account to create, and according to the available information no personal details are requested. It is optimized for phone screens, so most people finish it on a smartphone in a few minutes and land on a result screen they can read, screenshot or share. Desktop browsers work just as smoothly if you prefer a larger screen.
It is worth being clear about what this is not. The model names that appear as results are labels, not measurements. Landing on GLM-5.3 does not mean anyone tested your abilities, and it does not mean GLM-5.3 outranks other models in any real benchmark. Treat the quiz the way you would treat any lighthearted internet test: a playful mirror that is fun to argue about, not a scientific assessment of you or of the models named.
- Free and browser-based
No download, no payment and no sign-up. You open the page, answer the questions and read your result.
- Built by Lattice
The developer Lattice released it on October 18, 2026, and it spread fastest among Japanese-speaking AI fans.
- Model names as labels
GPT, Claude, GLM and Qwen appear as result categories, each with a short persona description attached.
- A few minutes, phone-friendly
The layout is tuned for mobile browsers, so a complete run takes only a few minutes from start to result.
How the Quiz Turns Your Answers Into a Model Label
The quiz follows the familiar rhythm of a personality test. You are shown a series of statements or multiple-choice questions about how you think, work and talk, and you pick the option that fits you best. Each answer nudges you toward one or more model categories. There is no timer and no right answer, so the honest choice usually produces the most recognizable and most entertaining result.
Behind the scenes, the scoring is a mapping exercise rather than a psychological instrument. Answers are tallied against the traits associated with each model's public reputation: one model reads as the careful analyst, another as the fast generalist, another as the blunt specialist. Whichever category collects the strongest signal becomes your label. The quiz does not publish its exact weighting, so treat any claim about a precise algorithm as unconfirmed and check the official site if you are curious.
Your result arrives as a model name plus a one-line persona description. The example that circulated most widely was GLM-5.3 paired with a description calling the person a dangerously sharp specialist, which is flattering, slightly ominous and very quotable. That tone is the point: the descriptions are written to be shared, and the fun comes from reading a friend's label and deciding whether it fits them better than it fits you.
- Answer honestly, not strategically
Picking the impressive-sounding option tends to flatten your result. Choose what you actually do and the label usually feels more accurate.
- One label, one description
You receive a model name and a short persona line, which is the part people quote when they compare results with each other.
- No timer, no login
You can take your time, and the quiz does not ask for an account or personal information before showing your result.
- Scoring details are not published
The exact weights behind each label are not documented, so treat the mechanism as a fun heuristic rather than a formula.
Reading, Sharing and Comparing Your Result Without Overthinking It
The healthiest way to use the quiz is as a conversation starter. Take a screenshot of your result screen, send it to a group chat and ask everyone else to run it too. The interesting part is not the label itself but the argument that follows: why did the calm planner get the fast generalist, and why did the loudest person in the room end up with the careful analyst? That mismatch is where the entertainment lives.
Be careful about how far you push the comparison. A quiz label is not a personality diagnosis, a hiring signal or evidence about which AI model is best. If you are genuinely curious about model quality, look at published evaluations and hands-on testing rather than a browser quiz. And if you share your result, remember that the persona text is written to be flattering and dramatic. It is a joke with a straight face, not a compliment you earned.
Retaking the quiz is allowed and mildly revealing. If you answer the same way twice and get different labels, that tells you the mapping is loose and the categories overlap. If you answer differently and get a different label, you have simply described a different mood. Either way, do not read deep meaning into it. The quiz is designed for a few minutes of amusement, and it delivers that well.
- Share the label, not a verdict
Sharing your model name is fun; presenting it as proof of anything is where the quiz stops being fun.
- Compare with friends
The quiz is at its best when several people run it and read their persona lines side by side.
- Retake it if you like
Nothing stops you from running it again, and inconsistent results simply show how loose the categories are.
- Keep it separate from real model news
A result like GLM-5.3 is a label, not a product announcement or a performance ranking of any kind.
How It Compares With Other Quizzes and Who Should Bother
Compared with classic personality quizzes, this one swaps psychology vocabulary for AI vocabulary. Instead of introvert and extrovert, you get model names that carry their own reputations in tech circles. Compared with actual model leaderboards, it is doing something completely different: it measures nothing about the models and everything about how you answered a handful of questions. That gap is exactly why it works as entertainment and fails as evidence.
Its limitations are straightforward. The result set is limited to the models the creator chose to include, the descriptions are subjective, and the scoring is not published. Availability is simple but worth checking: the quiz is free and playable in a browser at ai.tatoe.top, and the available information describes it as mobile-friendly with no login. If the page changes or adds features, the official site is the only reliable source.
Who is it for? Anyone who enjoys personality tests, anyone who works with AI tools daily and wants a laugh, and any group chat that needs a five-minute icebreaker. It is not for people looking for a serious self-assessment, and it is not a way to decide which model to use for real work. Go in expecting a joke with a well-written punchline and you will get exactly that.
- Different from a leaderboard
Model rankings measure model performance; this quiz measures nothing and labels you instead.
- Different from a psych test
It borrows the format of a personality quiz but makes no clinical or diagnostic claims about anyone.
- Best in a group
The quiz is most enjoyable when several people compare their persona lines and disagree about the results.
- Check the official site for changes
Question counts, features and the available result set can change, so ai.tatoe.top is the place to confirm details.
Step-by-Step Guide
Open the quiz in your browser
Go to ai.tatoe.top on your phone or computer. The page loads directly with no download, no account and no payment, and the layout adapts to small screens without any setup on your part.
Answer the questions honestly
Work through the questions at your own pace and pick the option that matches what you actually do, not what sounds most impressive. There is no timer, so there is no reason to rush through the choices.
Read your model label and persona line
Your result names a model such as GPT, Claude, GLM or Qwen and adds a short description of the kind of person you supposedly are. Read that description as a joke rather than a verdict about your character.
Share it and compare
Screenshot the result and send it to friends, then ask them to run the quiz too. Comparing labels and arguing about who got the wrong one is the real entertainment here, so keep the tone light.
Retake it or move on
If you want a different label, run the quiz again with fresh answers and see what changes. If you want real information about AI models, look at published evaluations and hands-on testing instead.
Frequently Asked Questions
Is Which AI Model Are You free to play?
Yes. The quiz is free and runs in a web browser at ai.tatoe.top, with no login or personal information required according to the available information. You do not need to install an app or create an account. If anything about access or pricing changes, the official site is the place to confirm it.
Does a result like GLM-5.3 mean a new AI model was released?
No. The model names in the quiz are result labels, not product announcements or performance rankings. Getting GLM-5.3 says nothing about that model's real capabilities and nothing about yours. It is a playful comparison, so keep it separate from actual AI news and benchmark results.
How long does the quiz take, and does it work on phones?
It is built for mobile browsers and takes only a few minutes from the first question to the result screen. Desktop browsers work too. There is no timer, so you can take longer if you want, and you can stop partway without losing anything important.
How is the result calculated?
Answers are matched against the traits associated with each model's public reputation, and the strongest match becomes your label. The exact weighting is not published, so treat any detailed explanation of the algorithm as unconfirmed. The quiz is designed for fun rather than scientific accuracy.
Can I share my result with friends?
Yes, and sharing is a big part of why people enjoy it. Screenshot the model name and persona line and compare with friends who take the quiz too. Just present it as a joke rather than a judgment about you or about the models named.
