B

building

[B @BlakeStone]

@ziwenxu_ Idk I would argue the opposite direction.

Fable is the best creative and codes better for me than anything.

[Ryan Sayles @RyanSayles_]

I’m sure everyone will enjoy the moment that changed Mississippi forever!! @scott_colom I guess Mississippi Matters now huh?? 😂🔥

[B @BlakeStone]

@ziwenxu_ I'm gonna main projects moving forward... This is awesome

[B @BlakeStone]

@ziwenxu_ Gonna have to see how well they handoff projects to other accts lol

[Ziwen @ziwenxu_]

Claude code projects is kinda insane

> give it one task
> it splits into a bunch of agents running at once
> the coordinator tells you what each one is doing
> custom themes so it finally looks good lol

Our agents with a manager now

[Aaron Ladd @aaronladdtv]

Riding out the groundstop at the DFW Chili’s. Ask me anything

[Forward Spin @forwardspin111]

Pick another pronoun? She/it

[Ethan Brooks @alt_w_v_g]

Just let me know which one I should restart

[B @BlakeStone]

Im saying that proportional to total wages, taxpayers paid a larger percent of their income during this period, leading to rampant economic growth and services development.

Now, people pay a smaller proportion of their wages. It doesn't matter if its more dollars or even a different value relative to, its the proportional amount paid that matters.

And if youre saying we shouldn't have higher taxes because Amazon exists, I would argue the exact opposite. They should be hiring BECAUSE Amazon exists. Lmao

[B @BlakeStone]

@jacl0e @NadaKaupas @bennjordan Proportionally, honey.

[B @BlakeStone]

@jacl0e @NadaKaupas @bennjordan But they were paying more then, than now, yes?

[will depue @willdepue]

I have a major result to share.

I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China.

This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model.

Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves.

I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems.

I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used.

It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20.

These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs.

I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at https://t.co/YOtzND0Jqk.

There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.

[B @BlakeStone]

@TheCryptoData @allietheicon Join the club cryptodata, I have 7 🤝 please send help 😭🫪

[Acyn @Acyn]

Talarico: On election night, as I was delivering my victory speech, I noticed a woman up front. The lights were bright, so I couldn’t see her face, but I was struck by her attention, how closely she was listening, how much this moment meant to her.

As I finished speaking, I looked down again, and she had made her way to the front of the stage. I bent down to shake her hand, and I realized who she was.

She looked me in the eyes, and she said, “I’m so glad you have my father’s job.”

Lucy Baines Johnson was there to see another public school teacher take her dad’s Senate seat.

[vio 🖤✨@ TWITCHCON @viotubing]

bro 😭😭😭

[B @BlakeStone]

@arshitamisr @Ritsrroma Oh, no, you seem 100% real.

[B @BlakeStone]

us-east-1 flashbacks

[B @BlakeStone]

@arshitamisr @Ritsrroma Investigated a bit, that is just a bot that someone has doing AI on other peoples posts and reposting them.

Look at the words on the screen in their post. They ran both images through AI but only needed to run the first one to swap the person. 100% AI.

[B @BlakeStone]

@arshitamisr @Ritsrroma And yours isn't AI'd to hell