How we picked 14 from 100+ smart glasses
When we started this store, the question wasn't "what should we
sell?" but "what shouldn't we?"
The smart-glasses market in 2026 looks something like this:
roughly 100 models, made by roughly 30 manufacturers, sold under
roughly 200 brand names. The same physical glasses ship with
different logos to AliExpress, Amazon, niche e-commerce stores
across Europe and North America, often at wildly different prices.
The temptation, as a small dropship store, is to list everything.
List 200 SKUs, take a 15% margin on whatever sells, optimize for
SEO. We know stores that do this. They make money.
We chose not to.
Here's what we did instead:
1. **We surveyed the supplier market**, not the consumer one.
That meant working directly with TVCMall and a handful of
factory contacts in Shenzhen, asking what's actually shipping
in volume and what's vapor.
2. **We bought samples.** All 14 models we now sell, plus another
dozen we rejected. We wore them. We translated their apps. We
broke two of them on purpose to see how returns would work.
3. **We compared them to the rest.** Our comparison table tracks
56 models including all the big brands (Ray-Ban Meta, Amazon
Echo Frames, the new XReal lineup, Bose). The ones we sell are
flagged, but we deliberately show models we don't.
4. **We priced them honestly.** Our prices are roughly 3× factory
cost — covers shipping from Shenzhen, EU VAT, customer service,
and a real margin. Nothing exotic. Some models carry a launch
discount that fades into the regular price over the next quarter.
What you see on this site is the output of that filter. Fourteen
models that we'd actually recommend to a friend.
The next question we get is: *why so few?*
The honest answer: more than fourteen and we'd be guessing. With
fourteen, we know each pair. We can tell you which one fits a
narrow face, which one's terrible at translating Spanish, which
one's actually water-resistant vs. just marketed as such.
If you're picking one and want help, try the
[Configurator](/pages/konfigurator) — five questions, three
recommendations, all 56 models considered, the ones we stock
flagged.