Every hyperscaler hosting these models outside of FEDRAMP environments has been compromised by every regional power’s intelligence services. Fable was running all over the world until today.

AWS and friends are very good at providing excellent enterprise grade security, but it’s literal child’s play for nation state threat actors to exfil these models.

TEMPEST / EMSEC alone is a wide open door for unclassified datacenters when the Mossad’s out to get you.

But you can see this is not true (yet); competitors/Chinese labs are less than 6 months behind: either via leaks or by just stumbling on the same improvements with time/effort.

The gamers would really be complaining about why they can’t run Fable.torrent on their gaming PCs

As opposed to what, the US military, or better yet Israel (because we all know they won't be excluded) using that model to drive weaponry that kills people?

Your hypothetical implies that there is a better alternative, but when those models are "restricted", in practice that means that the only people who have access to them are precisely those who can and will use them for the worst kind of shit. So yes, releasing them to the public is a better deal, ethically speaking, at least then the playing field will be slightly more equal.

What if I told you there are no safety guardrails. I used GLM 5.1 and had fable literally build a harness to avoid triggering guard rails. I built skills carefully and had Fable doing vuln research and exploit repro in a few hours. I called the project manhattan. The GLM models are down for almost anything so I named it Oppenheimer. It orchestrated the fable CLI agents via tmux. This whole Fable/Mythos thing is such a fucking joke. It is all PR and theatre and they know it.

> Very few businesses can pay 1M without blinking. None of my customers.

Then they won't survive the termination boundary.

Too bad. Should have had more cash.

The biggest open models are also Chinese

how is that different from US AI that self censors and is banned from release by their emperors?

Deepseek's base models aren't censored.

That's the fundamental difference with open models, anybody can run and tune them any way they like. The real difference in philosophy is that Americans companies treat the model as the product, while Chinese companies see models at infrastructure you build products on top of. You amortize the cost of deploying it at scale by sharing knowledge and iterating quickly to bring the cost down.

I don’t need an AI to tell me about Tiananmen Square. I need it to do boring grunt work.