feels like the western world has transactionalism the wrong way round. buying a coffee comes with compulsory small talk, but helping a neighbour needs a reason beyond just being neighbours.
5
289
more people use a product, the company grows, and the product starts reflecting the org structure.
people move to a smaller, better built alternative, until it starts looking like its org chart too.
i call this conway cycle
1
2
274
Halil Cetiner retweeted
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 unemploymentlabs.com/checkpo….
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.
introducing boxrouter, the openrouter for sandboxes.
one api. every sandbox provider. automatic routing based on price, cold starts and whoever raised a seed round this morning.
the sandbox landscape moves fast. we handle the complexity so you can focus on shipping.
Sandbox provider are trending now, so here is a list of them ordered from most expensive to cheapest
(All prices are for 2 vCPU / 4 GB, running for one hour).
27. Depot — $0.60/hr
26. Vercel Sandbox — $0.341/hr *
25. Fly.io Sprites — $0.315/hr *
24. AWS Lambda MicroVMs — $0.252/hr
23. Tenki — ~$0.24/hr (its CI rate)
22. Modal — $0.238/hr
21. Upstash Box — $0.20/hr *
20. Cloudflare Sandbox — $0.180/hr *
19. Namespace — ~$0.18/hr (rough)
18. E2B — $0.166/hr
17. Daytona — $0.166/hr
16. Thunder Compute — $0.166/hr
15. BoxLite Cloud — $0.158/hr
14. Dedalus — $0.149/hr
13. Freestyle (YC S24) — $0.132/hr
12. CreateOS — $0.119/hr
11. exe.dev — $0.105/hr
10. AWS Nitro Sandboxes, on-demand — ~$0.083/hr
9. Northflank — $0.067/hr
8. AWS Nitro Sandboxes, 1-yr Savings Plan — ~$0.059/hr
7. Quivvo — $0.054/hr
6. Celesto — ~$0.030/hr (its CI rate)
5. AWS Nitro Sandboxes, Spot — ~$0.029/hr
4. Boat (YC F26) — $0.018/hr
3. InstaBox (YC P26) — $0.017/hr
2. Cube — ~$0.016/hr (shared vCPUs)
1. Agent 37 (YC F26) — ~$0.006/hr (shared host)
4
12
748
not sure teaching my daughter to celebrate failure this extravagantly is the best parenting strategy, but it was fun
6
399
accidentally brought an empty snack container back from Turkey with some rubbish stuffed inside. luckily, the UK will probably send it back to "get recycled".
4
386
Extreme Intelligence (Xi)
Many people think that the words “Artificial Intelligence” are inaccurate, and very ineloquent, relative to AI, or Artificial Intelligence. A far more elegant and accurate description of this new phenomena would be Superior Intelligence (SI) or, Extreme Intelligence (EI) or, Supreme Intelligence (SI). This is a Poll, and I would appreciate everybody voting! Which is the best name for this ever growing “Revolution?” President DONALD J. TRUMP
2
3
416
I've been using Codex roughly this way for a while and can't recommend it enough. I have just a handful of pinned orchestrator threads, plus a standalone app for tracking the subthreads they spin off.
Grok Bot is so much directionally better than everything before it
Feels like most harnesses, even coding harness, should adopt the patterns they pioneered
I don't want threads, I want topic areas with a main agent that orchestrates the work for those areas
2
538
Really disappointed with my first experience with @RegusGlobal in Turkiye.
What I thought would be a simple one-month office rental has ended up costing me well over twice what I originally expected, thanks to unexpected extra charges and an additional month I never wanted.
every time i order an americano and get asked "do you want milk in it?" or "is that black?", a small part of me dies
2
5
587
ordering a simple black coffee turning into a half-page dialogue every time drains my energy
- hello
+ hi how are you
- i’m fine thanks
+ it’s a lovely day out there
- yeah it is
+ what can i get for you
- can i get an iced americano please
+ regular or large
- large please
+ do you want milk in it
- no, just black
+ is that all
- yes thanks
+ it’ll be £4.50. tap whenever you’re ready
- *taps phone*
+ it’ll be ready on the other side
208
opened @LinkedIn before coffee and got told i have more than 30 years of career left. brutal way to start the day
1
476













