My doubt in the factualness of this claim is too high to allow me to be interested at all. Especially with how bad my Claude Code experience has gotten in the past weeks. I thought it was me until I started reading other people's accounts of how it seems to have gotten "lazy" and "dumber" since around February. If they have such a capable aka dangerous model, how is no residual benefit showing up in the currently "in the wild" models? I know my anecdote is not proof of decline but I can only speak from my own experience(s).
This comes up from time to time and although my experience is anecdotal, I see clear degradation of output when I run heavy loads (100s of batched/chunked requests, via an automated pipeline) and sometimes the difference in quality is absolutely laughable in how poor it is. This gets worse for me as I get closer to my (hourly, weekly) limits. I am Claude Max subscriber. There’s some shady stuff going on in the background, for sure, from my perspective and experience during my year or so of intense usage.