Samuel Cardillo

direct kinetic impact. a flying sword. 450km/h. updated video showing exactly that. we're also working on the explosive variant. only for authorized partners. dms are open.

reject robots, reject tractors, reject shoelaces, reject cars, reject the internet, reject the wheel, reject the radio minecraft was cool tho.

@NicoGrassetto @amjadt25 saudi arabia 🤷‍♂️

@CardilloSamuel @amjadt25 So he got cucked at the airport by the IRGC and there were no diplomatic consequences? 😂 I thought this only happened in with EU diplomats

The former Saudi ambassador to Iran says the IRGC gave him four slaps. His response? “I didn't do anything.” Four slaps and bro activated airplane mode.

@camski its probably going to escalate but probably not as much as 1st phase of the war. its chill.

@nawafsama_ wafi <3

Week off to relax

@MLK__TV @msuiche gratuit, my treat

@CardilloSamuel @msuiche Je te dois combien pour le pexing sur WoW ?

after talking with @msuiche about how rude it is for people to come ask you for advices and then literally not apply them, i've decided to paywall my time because i do believe that when you pay for something, you appreciate the value of it more. so if you want to ask me advices abour your business, i am more than happy to provide, for 199 USD. calendly.com/iam-givingyoum…

@msuiche @DJLougen @neutralityorg based

@DJLougen @CardilloSamuel @neutralityorg stay neutral

@Bouazizalex Gamers ftw

People always find it crazy when they learn that Deel is a 7,000+ person, fully remote company. They always ask me, “How do you guys do it? Doesn’t it feel weird never meeting your team? I could never.” Funnily enough, my answer is always the same. It’s all I’ve ever known. It feels natural to me, and I wouldn’t do it any other way. I do think it’s generational, though. At least for me, it started young. When I was 12, I joined an AoE3 clan with people I’d never met. Every day after school I’d rush home to jump on TeamSpeak, chat for hours, test crazy new strats, and compete. Somewhere along the way, they stopped feeling like strangers. Years later, League of Legends was no different. Grinding ranked, inviting a random teammate to duo after a great game, and before you knew it, you were playing together every night. Looking back, I think gaming taught an entire generation how to collaborate remotely long before Slack, Zoom etc existed. It normalized building trust, friendships, and shared goals with people you might never meet in person. I think that’s part of why remote work has always felt so natural to me. For us, collaborating and trusting through a screen was never the new thing. It was where we grew up.

@4484 install a local agent locally, rerun the same task, see if it works

@CardilloSamuel running this complex inbox triage agent for a client (lots of RFQs, quoting etc..) and switching to gemini for cost cutting made the decisions retarded. sonnet 5 does ok. frontier just hit everything out of the park.

you do not need frontier ai for everything you do. as a consumer, sure, have fun. but as a business? not really. a marketing department, an accounting team, … they do not need fable. what they need is availability, trust in the weight being used (frontier lab quantize/change the weight continuously to mitigate peak hour usage), trust that the model they’re using isn’t going to be deprecated in few months. and on top of that, token control. as a business, if you pay api usage, you literally can’t predict your bill. it might go up, it might go down. sure you can limit your employees usage but you don’t control the api pricing. so the only way you they make money is by making their customers pay for the tokens. that’s wrong. local ai fix all of this.

@Helios_Rises @msuiche that username 😂

@CardilloSamuel @msuiche Dont listen to this guy. I'll give you advices for free.

@4484 i don’t know, I don’t really use comparison this way - and there is already tons of available benchmarks. what i care about is if the model can do the job it is required to do. and qwen 3.6 can do most, if not all, things that businesses i helped have had needed to do.

@CardilloSamuel where do you rate latest qwen in comparison to closed frontier labs? sonnet?

i keep criticizing europe - for many reasons, including the fact that my passport renewal has been ongoing for over a month now. but i secretly want europe to succeed. i just know it can't. not in the current state it is. and i know there isn't much i can do. the european union is a corrupted organism that has transformed entire countries into bananas republic. globalist have taken over governance with regulations being used to control population. maybe one day. maybe the population will wake up and reshape the absolute joke the european members have became. starting by removing the lobbyists out of schuman.

@giovanniJurado @msuiche i still post online, sharing my findings and answering tweets. i just wont' advice a business/entrepreneur for free anymore.

@CardilloSamuel @msuiche Free advice for the proletariat such as myself

@giovanniJurado @msuiche what was?

@CardilloSamuel @msuiche Well... It was good while it lasted 😭

@Almutairi_ae @msuiche i've tried the simple and genuine but 99% of the time, people just value it. no more free - both for them and myself. if i know they paid for my advises, i'll provide even better advise. and if they know they paid, they'll listen more.

@CardilloSamuel @msuiche Id tier it simple geniune is free, anything that requires time and effort $$$$$

Today we present Morpheus, a persistent enterprise simulation platform designed to make Continual Learning a reality. Morpheus is the world’s first real world Reinforcement Learning environment. Every Reinforcement Learning environment operates in the game world. Benchmarks like Atari, OpenAI Gym, MuJoCo, and Procgen are all small, game-like worlds that reset every few minutes. But the real world never resets. A business keeps running and evolving everyday. We tested how frontier LLMs would perform in realistic and dynamic business environments 🧬on Morpheus. The main conclusion was that LLMs are not continual learners. 🧵Here’s how we did it and what we learned:

@brivael this is the right way of doing things.

Les seules réunions qu’on a chez Argil, ce sont des one-on-ones de 5 minutes maximum avec chaque membre de l’équipe, avec qui je me synchronise tous les lundis. Tout le reste se fait de manière organique. Et si on organise une réunion dans laquelle une personne se sent inutile, elle a le droit de partir en disant : « Je ne sers à rien dans cette réunion. » Rester serait plutôt un signe de faiblesse.

@NicoGrassetto @msuiche 100%

@CardilloSamuel @msuiche basic human psychology, you should have done it from the get go haha

the only thing that doesn't change is the local model you've deployed on your machine. that's one of the main thing that makes local ai so important. forget security, price, ... CONSISTENCY is what matter the most. no one to nerf nor cancel you.

@h1kz0r @Tech2Wild it is not about replacing frontier model,. x.com/CardilloSamuel…

@Tech2Wild Yeah no. The spark is great for MoEs and compute heavy workloads but let's not kid ourselves. No one is replacing a frontier subscription with any number of sparks. It's not built for it. The cope in X about this is just too much and very counter productive.

DGX Sparks will win the race. Models will only get better, more optimized, faster, smarter, etc. Just think about it, Deepseek v4 dropped and 2 Sparks were running it at 30 tok/s. DSpark came and not it is pushing 60 tok/s. It's only going to get BETTER and BETTER.