Question on when: https://manifold.markets/PrestonJensen/anthropic-announces-a-new-model-whe

Created a question about Anthropic announcing a new model today: https://manifold.markets/MalachiteEagle/will-anthropic-announce-a-new-model

https://x.com/theinformation/status/1890091003446718772

Dylan Patel repeats claim about Anthropic having a better reasoning model than o3: https://x.com/mark_k/status/1886769660344877073

@MalachiteEagle Sure, they have it, but I don’t think that’s evidence that they’ll release it. COT is expensive, and Patel has also said that Anthropic is fine keeping its big expensive models to themselves in order to accelerate their engineering work.

@cherrvak I think Anthropic cares about attracting talent, and it's valuable to have prestigious releases of frontier models towards this goal

@MalachiteEagle Ah, but as your linked tweet claims, they are “safetyists”, regularly putting out press about the dangers posed by models just slightly more advanced than whatever their API offers. And are they still focused on hiring at all costs, what with their fancy models improving the efficiency of the expensive engineers that they already employ?

@cherrvak I think Mark Krestchmann makes an interesting point but I think he is mostly wrong in his assessment of the situation.

they are edging us right now

Will hurt them a lot now if they don't release a reasoning model. Before deepseek ok, after deepseek not so ok