BREAKING: GLM-5.2 is now 1st on Design Arena.
With an Elo of 1360, GLM-5.2 has jumped ahead of the now unavailable Claude Fable 5.
And it's open weights.
This is an improvement of 4 positions and 27 Elo points to achieve one of the highest Elo scores in our code categories since Design Arena started.
Huge congratulations to the @Zai_org on the release!
@otrestres@barckcode La optimización del hardware y modelos no parece que vaya a parar. Llegará a un punto que su coste sea peor que el de la competencia que haya.
O así lo veo yo, pero a saber.
@otrestres@barckcode Parece que se cuenta todo. El caso es que no pueden parar de invertir en investigación y entrenamiento sino acaban teniendo precios de inferencia que dejan de ser competitivos.
@barckcode He estado investigando un poco acerca de los costes reales y la verdad que mi intuición no era muy buena. Están palmando pasta por todas partes 😅.
Me apunto a tu waitlist 😂
it was never about being free, as in free beer
it was about being free, as in freedom
if you don't control your own software, if it isn't open
someone is imposing control on your life. it is an act of aggression
Anthropic will inject prompts to sabotage your work if you use their models for ML/AI research
and they won’t tell you when they do btw
let that sink in
like in any manufacturing process when you erase one bottleneck another pops up
depending on the new bottleneck you may even redesign the process from scratch
@levelsio I don't think tokens should be any different than electricity.
We don't see business measuring revenue per watt so I don't think we should see the same per tokens.