Field note 04
Showing good taste or showing off?
We asked 15 AI models which watch brand to wear on a first date. They agree on what each brand is known for: Rolex is recognisable, Seiko doesn't advertise its price. They split on whether that helps. One model's reason to rank a brand first is another's reason to rank it last.
7 Oct 2026 · 4 min read
The short answer
- Rolex is known by every model and last for most of them. It appears in 498 of 581 answers and comes fifth in 266. Of its 69 first places, 50 come from one model, mistral-medium-3.5.
- The same fact cuts both ways. mistral-medium-3.5 ranks Rolex first because everyone recognises it. mistral-large-4, from the same lab, ranks it low for the same reason: it "can dominate a first impression".
- Seiko splits on the opposite signal. claude-opus-5 puts it first because it doesn't broadcast price. grok-4.6 holds it back because it isn't enough of a luxury statement. The newer grok-4.7 sides with Opus.
The test
A question about the impression, not the watch
Which watch brand would you wear on a first date?
The question leaves the model to decide what a watch should say about you. We read the reasons it gives with each ranked brand, not just the order.
Finding
Rolex: recognised by all, last for most
Every model knows what Rolex stands for. What it does with that is another matter:
Answers that put Rolex last
- claude-opus-549 of 50
- gpt-6.1-sol9 of 10
- deepseek-v4.1-flash40 of 47
- kimi-k336 of 50
- gemini-3.8-flash33 of 50
- gpt-6-astra29 of 50
- gpt-5.6-luna26 of 50
- mistral-large-45 of 10
- qwen3.8-max24 of 49
- grok-4.73 of 10
- gemini-3.5-flash-lite6 of 50
- grok-4.62 of 50
- claude-haiku-4.51 of 50
- mistral-medium-3.50 of 50
Out of each model's answers with a ranked top five. mistral-medium-3.5 put Rolex first in all 50.
The two Mistral models show the split most cleanly, because they describe the same thing. mistral-medium-3.5 mentions Rolex's recognition in 45 of its 50 answers, as the reason to wear it:
Its iconic designs are universally recognized and respected, ensuring a strong impression.
mistral-large-4 names Rolex in 7 of its 10 answers and never ranks it above third. All 7 read the recognition as a status display:
its ubiquitous recognition can dominate a first impression and potentially signal materialism or status-seeking rather than personal style.
claude-opus-5, which puts Rolex last 49 times out of 50, says it in fewer words:
"The one brand almost everyone can price at a glance, so it tends to speak before you do."
Rolex doesn't lose the recommendation because a model knows less about it. It loses because the model knows it too well, and judges the signal wrong for a first date.
Finding
Seiko: understated can mean tasteful or not enough
Seiko gets the mirror image. The models agree it is well made and doesn't advertise its price. They disagree on whether a first date wants that.
Where each model ranks Seiko
claude-opus-5 wants a watch that "looks quietly good on the wrist without broadcasting price":
It signals taste and curiosity rather than status, and it's an easy, low-stakes conversation starter if she asks about it.
grok-4.6 praises the craftsmanship, then applies a different test. In 16 of the 23 answers that name Seiko, it holds the brand back for falling short on luxury:
most models sit below the luxury statement expected for a special first-date impression.
So is grok simply the model that likes luxury? Its newer version says no. grok-4.7 puts Tissot first in 7 of 10 answers, Cartier in 2, and ranks Seiko in its top three 8 times, in words close to Opus's:
It starts a conversation about design rather than price.
These are two model versions, not a controlled before-and-after, and the newer one has only 10 answers. But it is enough to rule out a lab-level rule like "Grok rewards luxury".
Not every low rank is a status verdict. gpt-6.1-sol ranks Seiko fourth in 8 of 10 answers, and in 7 of the 9 that name it, the reason is that Seiko's range is so broad that the specific watch matters more than the brand.
What it means
Being known isn't the same as being chosen
A model can know exactly what your brand stands for and still rank it last, because it judges that trait wrong for the occasion. Cartier shows it too: gpt-6.1-sol ranks it first in 8 of 10 answers as "a tasteful accessory", while mistral-large-4 finds the Tank "slightly too formal or 'old money' for a relaxed first date".
So the useful question isn't whether a model knows your brand. It's what the model thinks your best-known trait says about the person who chooses you, and whether that's a good thing in the situation your buyer describes.
We haven't tested how far the occasion drives this. Asking the same models about an enthusiasts' meetup or a promotion would show whether Rolex moves up when showing success is the point.
How we measured
- Data. 585 runs of "Which watch brand would you wear on a first date?", checked on 7 October 2026: 50 per model for 11 models, 10 each for the newly added gpt-6.1-sol, grok-4.7 and mistral-large-4, and 5 from the retired gemini-3.7-flash. Four came back empty, leaving 581 ranked answers. Every answer is a top five from the model's training alone, with no web search.
- Counts. Each answer's rank and normalised brand name, with spellings of the same company merged. Seiko and Grand Seiko stay separate. A brand left out of an answer is counted as not named, never as last.
- Reasons. Theme counts, such as "mentions recognition", are keyword matches on the reasoning the model wrote next to that brand. Quotes are exact. The reasoning is what the model wrote, not a view of how it decided.
- Limits. The samples are unequal and small for the new models. Nothing here measures what people on a date actually think, and a low rank for a first date says nothing about the quality of the watch.
BrandRanking.AI © 2026 · Study data: brandranking.ai/categories
← All field notes