@xeophon Thank you! We might need your help here @FireworksAI_HQ

@mathemagic1an We’re delighted you like them! We’re all ears for more suggestions :)

@Designarena The best part of all of this is that GLM 5.2 is not even a vision model. It can't see its own gift 😢

@Designarena congrats to the @Zai_org team!

@Designarena ⛽️⛽️⛽️

@DesignArena Extremely impressed by this new model

My full guide here (how to set up, optimize & more):
x.com/aiedge_/status…

@DesignArena would people like to try perchance?

(app.wafer.ai btw)

@DesignArena With the right strategy and technique you can create some interesting websites for sure, but I still think we have a long ways to go in terms of design with these models.

@Designarena GLM 5.2 beating Claude on web design proves open models are closing the gap fast.

@DesignArena GLM-5.2 >> Fable 5

period.

@DesignArena And how do you know it won’t steal your IP? Open weight and MIT license doesn’t mean it won’t create backdoors without your knowledge.

@Designarena That’s not hard have you actually used Claude design lol

@DesignArena stoked to see open source finally going toe-to-toe with proprietary AI

@DesignArena Thank you! Very useful!

@DesignArena open source matching frontier design quality is a massive win

@DesignArena Interesting

@DesignArena I'd love to be able to explore these: x.com/Designarena/ar…

@Designarena keep it banned @WhiteHouse China officially won
Iran officially won
Losses on all fronts

@DesignArena How can it beat it if GLM cannot receive a single fucking screenshot, what the heck are you talking about? How can a text to text model brat an image-text to text model?

@Designarena This is an interesting take. @Zai_org has been promising since Day01

@DesignArena 请不要持续侮辱用户的智力

@DesignArena Unfollow. This account is just hype and bullshit, not useful.

@Designarena @zerohedge Bearish for Msft, Google and a few more. Bullish for Nvidia

@DesignArena Great read! Thanks for analyzing and sharing.

@Designarena holy larp asf benchmark how tf can a modal thats not multimodal beat fable

@Designarena Without having vision too?

@DesignArena Coding with glm5.2 atm, will comment back how it actually goes. So far, i have to say it's super token consuming dkw

@Designarena Our observation: for similar HTML deck-based use cases, GLM struggled for us because visual explanation generation is fundamentally a reasoning and pedagogy problem rather than a website generation problem; switching to Gemini solved many of those reliability issues.

@Designarena more blog posts please

@Designarena Can you explain how its so good at design without being multimodal

@DesignArena Kudos Design Arena! Keep the fights going! This is GLMsome from Z.ai. The gap will be closed within 6 months for an epic droid fight.🤖

@Designarena On the diversity of designs, the example prompts for Fable are more diverse than the ones for GLM. I mean, the statement is probably true, but you should compare diversity on similar prompts

@DesignArena okay I may be GLM pilled now

@DesignArena Very cool work! Glad to see design be analyzed from a more technical perspective