Anthropic’s Downward Spiral It’s no secret that Anthropic’s been crumpling under the load of the influx of new users. There have been complaints about outages and quality , its million-token context window for Sonnet 4 and 4.5 was dropped to 200k , and the company accidentally banned OpenClaw creator Peter Steinberger (less than two months after he joined OpenAI). And that’s just happened this month! After I wrote this post and scheduled it for Monday, I was updating my timeline and came across this post on Fortune referencing a postmortem from Anthropic where the company owned the diminished quality of Claude Code since early March. Fwiw, gG=iven their ethical stance on protecting Americans from illegal surveillance and not allowing their model to be used to power fully autonomous weapons—no matter how many tantrums Department of Defense Secretary Pete Hegseth throws —I feel a deep loyalty to Anthropic.…