Cafezin?

𝒢 ≔ π^(-1/2) ∫ℝ e⁻ˣ² dx = 1

[Cafezin? @davi_xch]

@GrlSemPaciencia @LeopoldinaDom Se o teu conceito de pagar as contas é o governo extorquir o nosso dinheiro e devolver 5% como se fosse caridade aí tamo aceitando a escravidão 2.0 de uma vez.

[Cafezin? @davi_xch]

@whyisjake

[Cafezin? @davi_xch]

@paula_t_rio @Maurormf Isso sempre foi óbvio o povo brasileiro que é otario mesmo

[Alexandre @alexandr3_x]

A esquerda : “Viva o sus”

A realidade: chamando você pra um exame depois de 2 anos morto

[Cafezin? @davi_xch]

@nunomiguelc You should go apply for permanent residency since you are so impressed by basic tourism, these people smh.

[Cafezin? @davi_xch]

@thsottiaux @thsottiaux

[Cafezin? @davi_xch]

@thsottiaux gpt-6-astra-fast:xhigh - turn 7 (21 tools) - running - $1.3030 [42 tok/s]

Fast mode on 42 tps what is going on Tibo...

[Cafezin? @davi_xch]

@bradthomasbrown @thsottiaux Well in theory a complex SWE solution that would take 80 hours with Astra now takes 10 hours, which would be huge and would beat opus 5.5 if the cost wasn’t out of this world, I guess only millionaires and OpenAI employees can fully enjoy this feature for now…

[Cafezin? @davi_xch]

@Micha1978976 @buildwithrajath I agree on that, and I can tell you they are having multiple problems with their SDK solutions (which provides us with the agent capabilities), the codex app seems to be a mess with artifacts, and more. They ignored all of that and released dots…

[Cafezin? @davi_xch]

@Micha1978976 @buildwithrajath Like literally do y’all think compute and energy is infinite? I work 24/7 on optimizing enterprise CPU’s for a living because without software optimizations the current load collapses any infrastructure…

[Cafezin? @davi_xch]

@Micha1978976 @buildwithrajath Asking for more compute is the most bot type of take in the AI world, with C2C optimizations and inference load-balancing and many other techniques to optimize the current compute makes your type of request literally contrary to the evolution. More compute = +energy = + cost

[Cafezin? @davi_xch]

@rezoundous Let’s fuck our coding models up for DOTS! We need our little agents with avatars and a messaging app ui, it is much more important to create another openclaw type of slop and shove it up to the normies than actually provide useful solutions and hardware so we can actually evolve.

[Cafezin? @davi_xch]

@rezoundous Yeah I honestly think I’m giving up on OpenAI, they where goin the right direction listening to actual SWE’s, now they acting like any other delusional CEO with AI

[Cafezin? @davi_xch]

@thsottiaux Nobody asked for this, this is not even a update it’s just a harness maintenance, if this is the 28 day run holy Anthropic can go on vacation

[Cafezin? @davi_xch]

@thsottiaux Even as a pro 500 user I must say that Anthropic is in the upper hand...we need a more optimized quality, efficiency and speed.

[leo 🐾 @synthwavedd]

As suspected, Anthropic's subscriptions offer over 5X the value of OpenAI's (vs the equivalent API pricing) when using mid-tier models

OpenAI offers very slightly more usage of Astra vs Anthropic's Fable 5.5 offering in Pro & Max 20x plans respectively

Choose wisely!

[Cafezin? @davi_xch]

@thsottiaux More efficient ultrafast finally most of my runs don't take hundreds of hours on complex problems

[Cafezin? @davi_xch]

@lukevans @thsottiaux Bruh....are you kidding me. Reset to 1%, did I do something wrong @thsottiaux?