12 comments

  • ppyyss8 57 minutes ago
    As a UX/UI designer in Korea, I love seeing related products being released. I hope they become even more advanced in the future.
  • coryvirok 19 hours ago
    This is really good! It would be really cool to somehow get human designs in the mix to see how the models compare. I bet there are curated design datasets with descriptions that you could pass to each of the models and then run voting as a "bonus" question (comparing the human and AI generated versions) after the normal genAI voting round.
    • grace77 19 hours ago
      wow this is a super interesting idea, and the team loves it — we'll fast follow-through and follow-up here when we add it, thanks for the suggestion!
    • debesyla 16 hours ago
      This would be extra interesting for unique designs - something more experimental, new. As as for now even when you ask AI to break all rules it still outputs standard BS.
  • muskmusk 16 hours ago
    This is a surprisingly good idea. The model vs model is fun, but not really that useful.

    But this could be a legitimate way to design apps in general if you could tell the models what you liked and didn't like.

    • grace77 16 hours ago
      yes! that is the hope — /play is our first attempt at building out utility, would love your feedback and will ship hard to make it happen!
  • jjani 6 hours ago
    How about adding "mobile"? A lot of the time models tend to default to designs that don't make sense on mobile, even when instructed to design it as such.
    • anonzzzies 3 hours ago
      Really? When I have a system prompt 'mobile-first design' it 100/100 works perfectly. What sort of things are you trying?
  • a2128 19 hours ago
    I tried the vote and both results always suck, there's no option to say neither are winners. Also it seems from the network tab you're sending 4 (or 5?) requests but only displaying the first two that respond, which biases it to the small models that respond more quickly which usually results in showing two bad results
    • grace77 19 hours ago
      Yes — great point. We originally waited for all model responses and randomized the vote order, but that made it a very bad user experience -- some models, especially open-source ones, took over 4 minutes to respond, leading to a high voter drop-off rate.

      To preserve the voter experience without introducing bias, our current approach waits for the slowest model within each binary comparison — so even if one model is faster, we don’t display until both are ready. You're right that this does introduce some bias for the two smallest models, and we'd love to hear suggestions for how to make this better!

      As for the 5th request: we actually kick off one reserve model alongside the four randomly selected for the tournament. This backup isn’t shown unless one of the four fails — it’s not the fastest or lowest-latency model, just a randomly selected fallback to keep the system robust without skewing results.

    • ethan_smith 19 hours ago
      Adding a "neither is good" option would improve data quality by preventing forced choices between two poor designs.
      • grxxxce 19 hours ago
        this is a great note — will be sure to add!
  • calcsam 12 hours ago
    interesting idea, this benchmark maps fairly closely to the types of output I typically ask LLMs to generate for me day-to-day
  • justusm 18 hours ago
    nice! Training models using reward signals for code correctness is obviously very common; I'm very curious to see how good things can get using a reward signal obtained from visual feedback
    • grace77 18 hours ago
      As are we, seems like the natural next step
  • paulirish 11 hours ago
    It would lend credibility to publish your system prompt.
  • iJohnDoe 13 hours ago
    Very cool! Can the code and design that is generated be used?
  • filipeisho 20 hours ago
    [dead]
  • butz 15 hours ago
    [flagged]
    • grace77 15 hours ago
      I wish—just added them back
  • adi_hn07 19 hours ago
    [flagged]