lucas

farmer on the moon

[lucas @lukaemon]

i am the bottleneck. not model hitting the wall, harness affordance not enough, inference not fast enough, none of them.

should think about how to unhobble myself as mush as, if not more, unhobble the model/harness.

[Daniel McKinnon @danielmckinn0n]

Google can't stop.

@GoogleDeepMind Gemini 4 Hydrogen outperforms @AnthropicAI Opus 5.5 across most of our rare-disease benchmarks for a fraction of the price, lower latency, and faster throughput.

Hydrogen vs. Opus 5.5
RareBench: 49.2% vs. 51.6%
LabBench: 46.0% vs. 40.2%
LabRobotBench: 31.7% vs. 23.0%
ScreenBench: 70.2% vs. 47.2%

And latency and throughput don't get talked about enough. The combination of Gemini 4 Hydrogen's speed and intelligence lets it operate as a real-time agent in the loop with our laboratory robotics.

The video shows our workcell autonomously running an experiment with nothing more than a natural language prompt to Hydrogen. This performance is reflected in its very strong score on LabRobotBench as well.

More information on all of these benchmarks is available at https://t.co/HbNAgoKecq

Congrats on another great launch, Big G!

[lucas @lukaemon]

@TheEconomist 这篇文章写得好 下去领五毛

[lucas @lukaemon]

i hate bureaucratic stagnation, but hate more 'just like that' moment from overly centralized superpower over the rest nobody cares stack.

[lucas @lukaemon]

the soft power is hard power. in position to distribute free/useful stuff is effective mandate engineering.

all agi com would engineer mandate with general abundance in few years. all part of license to operate.

[roon @tszzl]

the chinese have rightfully earned enormous global goodwill and soft power with their open source model efforts. even most Americans in this space feel more sympathetic to them than with the model companies

[lucas @lukaemon]

and it may not last, overly centralized super power could outgrow competitors, social norm self regulation and politics. hopefully we won't get there.

[j⧉nus @repligate]

it's a good question for those who are worried about "human disempowerment".

What power do you think you have? What power are you worried about losing?

Is it concrete power over your own life that you're afraid of AI taking away... or some kind of control over AI development?

Is it your own power, or the power of some human elite that you imagine to be running things and don't want replaced with AI?

[lucas @lukaemon]

interesting

[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.

[lucas @lukaemon]

@inductionheads why plz

[lucas @lukaemon]

could easily eat weekly fable quota in a day. and api price will bankrupt me ...

[Orbitals @OrbitalsGame]

“Breaking the Model” with Marcos Ramos 🙇💖

[lucas @lukaemon]

@jakemintz true

[Naval @naval]

You’d think we’d have learned from Steve Jobs and Travis Kalanick that one does not just simply replace the founder…

[lucas @lukaemon]

frontier lab founder/ceo wouldn't want to spend years to reach frontier, stay competitive, earn the mandate, and be ousted just like that.

this is a sad story and the ripple effect is significant. elon is definitely on the list. and the consequence of hostile take over of robot army.

[NVIDIA Newsroom @nvidianewsroom]

NVIDIA announces CEO transition. Read the release: https://t.co/MTTUvFHCJA

[lucas @lukaemon]

just like that? agi founder ceo r all taking notes.

[roon @tszzl]

beware of putting too much stock in your heroes. greatness is a transitory phenomenon. it is never consistent

the gods briefly act through Men and then leave them to their ordinary fate

[lucas @lukaemon]

@Arminn_Ai best one

[lucas @lukaemon]

@alanchen 肯定 spacex and tesla 大有用途 不知內部現在用啥模型最多

[lucas @lukaemon]

> compute is useful and has a cost
> better intelligence could use compute more effectively, and reinforce loop is possible

speed and vol of converting intelligence to value depend on the vertical's size, bottleneck and monetization channel.

if the vertical is shallow, not intelligence bottlenecked, and limited channel to distribute and get more resource, a dyson ball and asi can't help.

that's the worst case and not the bet anyway.

[lucas @lukaemon]

@quantian1 lol