Sudo su

@cherry_mx_reds it crashes on small local models. useless.

Fun fact: many OpenClaw maintainers worked extremely hard to ensure OpenClaw runs well, even on a humble Raspberry Pi. Despite the rumors, and the constant reminder that if you don’t own 10 Mac minis you’re destined for the underclass, you don’t need a Mac mini to have a great experience with a personal AI assistant.

@d3layd let's goo

@sudoingX FINE FINE FINE YOU'VE CONVINCED ME TO LEAVE OPENCLAW

almost 3am anon, and before i crash, a reminder: if you're getting into local AI or agentic developments, hermes agent is the leanest door in. stands up in seconds, no onboarding maze. you just run it, and you're in. start with the thing that gets out of your way. night.

welcome herald. hermes agent everywhere.

@NeoAIForecast and you are neo the king maker

@sudoingX Let’s go. Sudo is king!

my name is sudo and i'm 26. i am going to build the biggest data centers in southeast asia. not to go chasing users. because the demand from what i'm building will get so big i'll have no choice but to own the metal myself. datacenters in southeast asia. then in space. remember the name. sudo.

it's a good day to drop that stinky lobster bloat anon.

@MikeBradleyAI good one. lol

@sudoingX Are you intentionally setting me up to share this 🤣❤️

i see it constantly now, everyone "making local AI the default," launching platforms to bring it to the masses. and i've yet to see a single one that actually solved the infra for real users. here's the irony nobody says out loud: almost every "local AI for everyone" tool still needs you to be a computer expert just to stand it up. you have to already know the thing it claims to remove the need for. that's not solving infra, that's a dev kit with better marketing. the actual problem is making local AI work for someone who isn't an engineer, is still wide open. i see it clearly.

@jenzhuscott followed. will check your work.

@sudoingX Call me if you need a e2e fully integrated manufacturing partner, from BESS to versatile advanced cooling, thermal reutilisation solutions! 🫶

@nickzilllla clean split

@sudoingX 2x rtx6000, 96gb. 128gb machine ram M3 ultra 256gb M5 pro, 64gb

okay nerds, how much memory do you actually own right now? not rented, owned, sitting on your desk. i'll start: > dgx spark, 128gb unified > strix halo, 128gb unified > 5090 laptop, 24gb vram + 64gb ram > 3090 node, 24gb vram + 32gb ram > 3060 node, 12gb vram + 16gb ram > old acer laptop, 8gb (yes it counts) > phone, 12gb ram 448gb of memory i own outright. all mine. flex yours.

@jtregunna ok you win this one comfortably.

@sudoingX 2.4TB system ram, mostly ddr4, and 132GB VRAM, but that's just what's running and known working, I own more

@FourEyedWiz appreciate you, watch what happens.

I don't know you, but yes, do it!

@realbrucemartin LET'S GO bruce.

@sudoingX took this advice from @sudoingX two weeks ago. Up 100 followers and gaining traction fast.

dear algo, somewhere out there is a founder about to quit x today. stuck under 200 followers for months, convinced the platform just isn't for people like him, that nobody's listening. show him this article. it's everything i learned going from there to 30k, it might be the thing that makes him stay one more week. and one more week is usually all it takes.

@Hikari_07_jp means a lot coming from you hikari. you're one of the few who gets why i keep these free.

@sudoingX I'm seriously grateful for this article.

@hughthebuilder because the sauce was never the secret, the execution is.

@sudoingX How is this sauce free bro

here is the quant cheat sheet nobody gives you straight. save this if you run local models: Q4_K_M , 4-bit. smallest, fastest, the one you use to fit a model on a card that "can't run it." real quality loss but usually fine. Q5 / Q6 , the middle ground. more vram, more quality. Q8_0 / FP8 , 8-bit, near lossless. basically full quality at half the size of bf16. the sweet spot when you have the room. bf16 / fp16 , full precision. the quality ceiling, double the size of Q8, the slowest. the rule: every step up in precision = better output but more vram and fewer tokens/sec, more bytes to move per token. what actually fits: 24GB card → Q4 of a ~35B, or full precision of a 7-9B 128GB box → bf16 of a 35B, or Q4 of a 235B i ran Ornith 35B both ways this week: Q4 hit ~78 tok/s, FP8 dropped the speed but the model visibly got smarter. precision isn't just cleaner text, it buys reasoning. pick your quant for the job, not the hype.

@gpt_alex yep, for real. xai dropped composer 2.5 into grok build, and premium+ / supergrok unlock it straight from the terminal. that's the cursor side of the spacex/xai merger showing up, composer's cursor's model, now it runs inside grok build too.

@sudoingX Grok premium plus gives composer 2.5????

if you're starting from zero and can only afford ONE subscription, don't spend it on chatgpt or claude. spend it on x premium+. it's the highest roi money i've ever put down, in order of magnitude. think about what x premium+ really is for someone young and building. the checkmark boosts everything you make, and once you're consistent and adding real value, the platform starts paying YOU. grok build comes bundled, which alone covers the cost. then there's the part nobody talks about, the audience you grow, the partnerships that find you, the paid work that comes from building in public. but here's the deepest insights almost nobody connects: you can auth x premium+ straight into hermes agent. that puts grok build, a frontier model, inside your own open agent, off the sub you already pay for. and hermes agent isn't a chat box, it runs a self improving loop and its own skills, so it gets sharper the more you use it. plug frontier intelligence into that and your machine stops feeling like a tool you open and starts feeling ambient, alive, an agent always running and improving alongside you. and it doesn't stop at grok, x premium+ comes with composer 2.5 too, whatever you point it at. one subscription that pays you back, builds for you, runs a living agent on your own machine, and opens doors the ai labs never will. if you're starting your journey, start here.

@devloperhs the sub alone doesn't auth it. run hermes model, pick the xpremium OAuth (SuperGrok / Premium+) option, and sign in right there. that OAuth step is what links premium+ into the agent. do it and it connects.

Nice break down , but still little skeptical about using grok build with X premium and not x premium +. Also verifying it with Hermes agent. I actually tried it and it tell me to buy the subscription , but I am already a premium member. So which premium member are we talking here , any resource I can refer ?

@MrSaneApps it's my daily driver.

@sudoingX This is not snarky I mean this as asked: What have you built with that setup?