I have no idea what Opus is outputting
Ok hear me out: I have a good level of proficiency in the English language and have been working as a software developer for almost a decade now. Since switching to Opus 5 I've noticed that...I have no clue what the model is outputting. The sentences are structured in a way that borderline make no sense at all. I don't know if it's linked to the eu ai act, but with Fable I have no problem at all! Compared to Fable, Opus generates so much bs filler commentary and keeps correcting itself over and over. I sometimes find myself reading the 5 pages of text produced for a css-token replacement 6 or 7 times before piping the output into Gemini to understand wft Opus is talking about.
Am i the only one? I'm working with caveman and the simple English skills enabled and use the wait-what skill when I'm about to give up.
How did you solve this issue?