@enheighted @aidan_mclau @Josh9817 R1 preview was out since November

@aidan_mclau @Josh9817 o3? just a month ago you had to fill out a form and have credentials to get access, pretty sure that was for safety reasons. If DeepSeek didn't release R1, o3-mini would still not be available to paying users. As usual you launched a bunch of shit at once to shadow competition.

@bycloudai Gemini is really not that good. It's cheap and has a big context, but it's dumb and hallucinates all the time

@MaxRovensky @kimmonismus yeah I tried it myself and it has to be something like that. these "safety" people are out of touch / psychos

@kennethnym @seatedro native or... tauri?

@berrybrazza sick, thank you

@0xluffyb @ludwigABAP deeper trvth seeker

@twobenjamins @yacineMTB Apple HomePod but with local inference hardware that can run 670B fp8 R1

@jarredsumner idk how you would do it but "use server" working with just `bun ./index.html` would be sick

@teortaxesTex @ThePrimeagen he's probably invested in nvidia because it's insane how he keeps repeating this to his fanboys

@wordgrammer @stefanboesen Slop take

@elder_plinius @astraiaintel @OpenAI Its not uncensored though, you didn’t give it an actual prompt that tests that

@ExaAILabs @D_Twitt3r Quantized?

@teortaxesTex it's 10x worse for image generation. The margins on LLMs are nothing compared to image gen

@ludwigABAP

@BretHatin @k_emhyr I know you're a subtweeting lil bich

@ZeitgeistPM @nearcyan no one knows you lil bro, „home grown” is not a flex in this context

@artiscoding @swlkr unc is yapping