I wrote an LLM inference engine from scratch in Rust.
No PyTorch. No ONNX. No Python.
Just one binary that downloads a model and runs GPU-native inference on Apple Silicon + NVIDIA.
It beats llama.cpp at Q4 on Apple Silicon. Here’s how it works 🧵
Sam Altman this week:
"We're going to see 10-person billion-dollar companies pretty soon."
"If I were 22 right now, I'd feel like the luckiest kid in history."
Most people will read that, feel inspired for three minutes, and go back to exactly what they were doing.
The ones who act will build an app studio this weekend.
One tool. Ten minutes to ship. A real path to $10K per month.
The window Sam is describing is open right now.
The people building inside it are not waiting for better tools, more time, or permission.
Here is exactly how to start this weekend.
Hey Ansem, here are my predictions that you bought AT LEAST one of these memes:
$WORLDCUP = a memecoin that is frontrunning the massive attention of the upcoming World Cup and is heavily backed by top KOL @metaversejoji.
$LMAO = a memecoin based on the classic "Laughing My Ass Off" internet meme. It boasts a strong, loyal community holding for months and offers real utility to boost trading volume via its in-house Volume Bot at https://t.co/a976MPVkLB. Heavily backed by top KOL @slingoorio.
$TROLL = a memecoin based on Trollface, one of the original and most iconic memes of all time on the internet. It’s reclaiming its $100m floor and is primed to pump toward $1B. Heavily backed by top KOL @CrashiusClay69.
$JOTCHUA = a fresh trending dog memecoin with a very strong and active community. It’s giving early $BILLY / $WIF vibes.
$NEET = a memecoin based on the "Not in Employment, Education or Training" lifestyle. It celebrates anti-hustle culture with the vibe “We bought a coin and never worked again.” It has a very strong and dedicated community, and is heavily backed by Mitch (sadly banned from X), who continues to support it strongly from his Telegram channel: https://t.co/c8HOhmBT3l.
$BUTTCOIN = a Solana memecoin created by Shytoshi (the same dev behind $FARTCOIN). It's a hilarious parody of Bitcoin, literally "Buttcoin" as the cheeky next Bitcoin. Pure meme energy, backed by many top KOLs and a very strong community.
$TRIPLET (Tung Tung Tung Sahur) = a viral memecoin exploding from the massive TikTok "Tung Tung Tung Sahur" brainrot trend. It’s riding huge TikTok virality with millions of views everyday and sounds spreading like wildfire. Backed by a lot of top KOLs and a strong community.
I hope @blknoiz06 that you're holding at least a few of the memes from this list, because they have the most potential in the current Solana market.
If not, FEEL FREE TO BUY, SHILL, HOLD, AND SEND THEM TO BILLIONS!!
The problem with $WORLDCUP here is not that the idea is bad
It’s that people have confused the size of the event with the longevity of the trade
The World Cup is massive, obviously..
But event coins don’t really trade the event itself. They trade the anticipation before the event
Moon landings, launches, listings, elections, sports events, it’s usually the same structure
Before the event, everything is still possible..
Maybe Brazil wins
Maybe the whole country coin meta goes insane
Maybe normies show up
Maybe every match creates a new pump
Maybe Elon tweets the opening ceremony
That is where the premium comes from, that is what gets priced leading to ATH
But once the World Cup actually started, maybe gets replaced by actual outcomes
And the issue with event coins is that the end date is visible to everyone
We all know when the music stops, so everyone starts thinking about when to leave the dance floor..
That’s why while the actual World Cup gets closer to its biggest moment, the on chain trade can be doing the opposite..
SpaceX just landed onchain.
The largest IPO in history, tokenized on day one. SPCXon is live on Ondo Global Markets across @solana, @ethereum, and @BNBCHAIN.
For the first time, a trillion-dollar IPO is onchain on the same day it goes public.
Elon Musk fell into depression as a teenager.
Nietzsche made it worse.
He tried to solve despair with philosophy.
It only trapped him deeper in his head.
Then imagination did what analysis couldn’t.
Here are Musk's secrets that will change how you see anxiety-forever: 🪡
1. You need a world bigger than your pain.
The part of SpaceX’s business that has short term upside is the terrestrial data center business. Right now they have 2 customers: Anthropic and Google. The fastest way to expand would be to acquire a world class data center developer like Cipher. The issue is it’s not for sale.
Fable 5 is state-of-the-art on nearly all tested benchmarks, with exceptional performance in software engineering, knowledge work, scientific research, and vision.
The longer and more complex the task, the larger Fable 5’s lead over our other models.
Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.
Its capabilities exceed those of any model we’ve ever made generally available.
The bottleneck in AI adoption was never the technology. It was the human hours to build and deploy at scale. Agents that build agents just made that problem disappear.
More from @btaylor and @Rivian's @RJScaringe on @CNN: https://t.co/vH69W0yBl4
I joined Sierra 6 weeks ago and very quickly thought “this feels different.”
To put things into perspective: in my last job, we had incredible AI tooling and effectively infinite tokens. But despite all of that, it often still felt surprisingly hard to translate capability into momentum.
Tokens were burned, code was generated, but progress still felt slower than expected.
I think I’m finally starting to understand why.
Now that engineering effort is no longer the primary bottleneck to producing output, many traditional software operating models start to feel much heavier:
- sprint rituals
- productivity proxy metrics (tickets closed, LOC, token usage)
- roadmap misalignment
- cross-functional dependency coordination
- layers of design + architecture reviews
- approval chains and stakeholder buy-in
- org boundary negotiations
In a world where it’s cheaper than ever to build and experiment, organizational latency becomes the real bottleneck.
What feels different at Sierra:
1. Quick feedback loops: We’re surrounded by opinionated people who aren’t afraid to say what they think. It’s often faster to iterate than debate. That forces sharper thinking and better design choices.
2. Agency: People are trusted to run with ideas. The distance between “this could exist” and “this is shipped” feels dramatically smaller.
3. Directional alignment: There seems to be a strong shared conviction around direction, which creates room for autonomy while still moving cohesively.
It’s still early for me here, but this is probably the clearest articulation I’ve found so far that “AI-enabled” and “AI-native” are very different things.
🚨@fomo founder reveals a surprising stat:
"90% of users on fomo today are not Crypto Twitter."
Out of 600,000 users, more than 500,000+ came from:
• TikTok
• YouTube
• Instagram
• Word of mouth
Many of them had never traded on-chain before.
The next wave of crypto adoption isn't coming from CT.
Se: "The only way crypto actually breaks into the traditional world is if you don't have to know it's crypto."
argue with your keyboard ↓