I've been working between Opus 5 and GLM 5.2 recently. I am finding GLM 5.2 wildly more effective than Opus 5 at finding and fixing bugs. YMMV, but was wondering if others had poor performance with Opus 5

Opus 5 has not gotten good reviews, although I haven’t ever used it

On the other hand, I understand it's worth upgrading to GLM 5.3

Yeah, let me upgrade our proxy models

My agent still uses habitually Opus 4.8 for main agent, and GPT-5.6-Sol for memory consolidation, DeepSeek-v4-Pro as workhorse for many well-scoped tasks. The main reason is what Graham says, people don't seem to have great experience with Opus 5 - you're not alone, Joe 😅

"Claudish", "Nopus" and "Dopus" are terms I've seen recently. Claudish being the language that Claude speaks, from and to which English must be translated.