I've been having fun throughout this debacle (posting femboy pics, indulging in heavy sarcasm), perhaps too much fun. Let me briefly spell out my object level opinion and go sleep.

First and most importantly, Anthropic is the #1 lab today. I say this without joy, my sympathies are obvious enough, but even on the American frontier I respect Gemini's long-term vision of minds and scientists embedded in generative worlds and OpenAI's reasoning fundamentalism more than this insipid, small-lab-smell agent-maxxing. Yet this is the reality. Anthropic has the strongest model on the market, and likely the strongest model period, and it's disproportionately better at accelerating ML research, which leads to a compounding advantage. They're wriggled through their once-crippling compute bottleneck. They lead in revenue, they have grotesque explosion of valuation, they have enormous and durable hold on the enterprise market, they are *this close* to SoTA on multimodality and scientific reasoning which were supposed to be strong suits of their competitors (even as they continue agent-maxxing), and they can serve this all *fast* and not even very expensively (Opus-fast tier prices, at about Opus-fast speeds). During early Mythos days, given rumors and leaks, it seemed like they just went all in scaling some uninspired architecture and OpenAI will easily exceed this level. But no. This is real. They might have figured something out indeed, some set of decision choices which have let them pack more capability into less. It's *not* a defensible moat, it's not anything OpenAI can't do. Bu it it s narrative victory, and in SF, narrative is often all that matters. OpenAI is obviously rattled. I don't want to point out examples, you should be able to notice on your own. They *did not* intend to allow Ant such an edge before an IPO. They're not just shitposting, this might be their biggest Code Red yet. We all should be rattled. Dario is trying to ride *three* tigers at once: AI takeoff and AGI/ASI control, societal and economic transformation resulting from it (with a transition to UBI nanny state via welfare funded by AGI, somehow), and the United States government. It's bold because it's desperate. He needs to position himself as a Patriot and Man with a Plan, because with the way things are going, one possibility is soft (or not even soft) nationalization, tiger #3 biting him in half after that welcoming nibble from the DoW. He tries hard to demonstrate that he and Anthropic are necessary for the System, equal stakeholders in the imagined global lib reich; and in doing so is also spreading the norms of Ant's behavior. Right now OAI's policy against using their models to train a competitor is very loose. What if Fable normalizes silent sabotage of, say, Docker? It's very close.

But Dario isn't stopping, he's giving detailed policy advice to the USG and beyond, he's preparing in his attempt to become too big to kill.

… not sure what else to say. Sounds obvious. This isn't the best timeline.

Again, I hope that we see a powerful 5.6 soon. Else Dario may ride his tigers. gn