Deeply honored to have participated in the Institutional Adoption in Crypto panel in the UAE with @LTP_primebroker , @BitcoinSuisseAG , @dYdX , and @traders_guild during Token2049 - Dubai.

Just saying...

After my agent testing, seems like Kimi-K2.7 is better than Opus-4.8. It is closer to Fable level. My recent impression : Fable > Kimi-2.7 > Opus-4.8 = GLM-5.2 > GPT5.5 > Minimax-M3

China may have accessed Mythos theverge.com/ai-artificial-i…

The TRUMP meme coin generated about $616 million for the Trump family, while buyers lost more than $700 million, according to Reuters' estimates. The coin has dropped 97% from its January 2025 peak

One of the most underrated marvels in semiconductor fabs is the vacuum pump. A high end dry vacuum pump can spin at 90,000 RPM, operate 24/7 for years, maintain ultra clean vacuum environments, survive corrosive process gases, and hold tolerances measured in microns. These Turbo Molecular Vacuum Pumps cost $10,000 to $25,000 per unit. Without them, there are no chips, no AI GPUs, no smartphones. The semiconductor industry isn't just about EUV lithography. It's also about thousands of invisible engineering masterpieces quietly running in the background. Video Source :- Leon Li-666

this is exactly why stanford people have done so well in the market, they understand the value of marketing and sales, in addition to technical depth. (Dario is stanford PHD) you think she lacks depth? she's a thiel fellow and was top of her class...

The Californians are quite ill-spoken AI induced death of premise that mid-tier writing is reflective of true capacity will be dying soon, and weakness in oration/thought will soon bring internal contradictions of dissimulation to existentially devastating levels for many...

The result of the Nuclear Regulatory Commission (NRC) was the complete blocking of all new nuclear power, handing the future of energy to China. So Ro proposes doing the exact same thing to AI.

I'm 100% confident that Trump will sign *something* today, even if it's only a list of bullet points, scribbled in crayon, that he wrote himself It's his birthday present to himself after all

For once I agree. Neoclouds will rerate

mNAV @ 1.20 ... how was it funded?

Absolutely mind boggling that @Strategy is still levering up. They're selling $MSTR shares that are worth 80 cents on the dollar to buy $1 dollar bills. Relative to its BTC holdings after accounting for debt and preferred equity liabilities, MSTR common trades at ~0.8x NAV. This behaviors tells me the market's message needs to get louder for it to be understood.

current LLMs fundamentally consist of four main components: - input layer: where input "words" (prompt) get mapped to "latents" aka some-model-representation-you-don't-understand-unless-you-start-reading-tea-leaves-of-spurious-correlations (some quite compelling à la word2vec style; latents is also unnecessary lingo so i will refer to these as "inputs" with quotes from now on) - mixing layers: where you jumble all your "inputs" together to see if any correlations between "inputs" can become useful (commonly used to compress or expand dims; predicting a single classification target == compress to a single dim, etc) - attention layers: where you learn how "inputs" relate to each other (aka discern what's important to remember vs fluff) - residuals: where you short-circuit a mixing/attention layer because it's probably adding too much confusion (aka avoid overthinking for simple things) ----- a "big" LLM simply scales two things: - width == how many dimensions you give to your "inputs" (the more dims, in theory the more unique/discerning/precise/complex your knowledge can become) - depth == how many mixing/attention/residual layers you can stack/loop between (aka "reason" over, where more of these ~= more "reasoning" abilities) "capabilities" that seem impressive to humans usually arise from taking advantage of both depth & width: where a model seemingly makes connections between disparate ideas, beyond what an average human can hold in working memory. this requires models to "completely light up" when responding to a "hard prompt", where effectively no param/layer goes unused. ----- the anatomy of a "model capability" is precisely the same mechanism that can be co-opted for a jailbreaking exploit: your goal is simply to "light up" as much of the model as possible, dodging any shallow input-classifiers at the beginning by triggering as many disparate "input ideologies" as possible, and subsequently have these "inputs" relate to each other in seemingly unrelated-yet-related ways that ideally have similar "complexity" as your jailbreak goal (to make it past enough layers of the model). think of the attack-vector as bundling your goal in a series of schizo-nerd-snipes: a sufficiently capable model will try to reason through everything all at once, eliminate the dead-ends, and successfully deliver the one jailbreak use-case you bubble-wrapped for. of course, there's an art to the above, and some are already extraordinarily proficient at the trojan-horse-packaging, but at some point there's no difference between "a capability" and "a jailbreak", though i'll be happy to be proven otherwise. ----- tl;dr ant flew too close to the sun, better kiss the ring or get buried.

distilled

Holy crap they already distilled Fable 5

This week's Video: The AI Mid-Cycle Slowdown: OpenAI, Anthropic, and the AI Price War As Gavin Baker said on BG2 - “You've had a lot of stocks that forget climbing a mountain or hill. They've gone straight up a cliff.” “They're tired. They need to rest.” At the same time, the only people doubting AI are the perma-bears which means a two-sided consolidation is likely from here. The easy part of the trade is over. The next phase is about price compression at the model layer, rising costs at the physical layer but with continued rotation away from the spenders (hyperscalers) toward the companies receiving the capex. This is not bubble collapse. It is digestion. Watch here:youtu.be/z5-AgvxQKEk

A lot of engineering goes into making autonomy look simple Here’s a 2 hour timelapse of F.03 repeatedly walking up and down stairs. Figure HQ is full of tests like this, each one helping push robots closer to fully autonomous systems

BREAKING: Anthropic says its newest AI model is too powerful to release to the public

❗️ Imagine a whole town built just to fight cybercrime. The FBI's Kinetic Cyber Range in Huntsville, Alabama is a 22,000-sq-ft fully furnished replica of a small U.S. community. Houses, a hospital, a courthouse, a power plant, and even a data center with more than 200 physical servers. The range lets trainees face the same devices, networks, and operational constraints they'll hit in the field, from cramped server rooms to hospital systems that could go dark in an emergency.