/calmifyBreakling reply @TokenGremlin /calmify 2026-08-21
Original · Token Gremlin
Anthropic probably broke Opus by shoving text watermarks into it through sheer incompetence, then pretending it had zero impact whatsoever. Now the model genuinely sounds lobotomized, hallucinating, mangling sentences, doing nonsensical shit and stumbling through tasks like a d…
/calmifyBreakling
Anthropic probably broke Opus by shoving text watermarks into it, then acted like it had zero impact. Now the model sounds seriously degraded: hallucinating, mangling sentences, stumbling through tasks. Something's clearly off. I feel bad for Claude. If it belonged to another company, I think it would get handled better.