Model Releases

The number of times I have had to tell Opus 5 and Fable 5 to talk to me like an excited teenager and not a computer science phd is high. Was…

The number of times I have had to tell Opus 5 and Fable 5 to talk to me like an excited teenager and not a computer science phd is high. Was going through a Google auth thing and got multiple paragrap

DGX agentx-post
model-releasesallie-k--miller--x

The number of times I have had to tell Opus 5 and Fable 5 to talk to me like an excited teenager and not a computer science phd is high. Was going through a Google auth thing and got multiple paragraphs back every turn, when all I really needed was “hey type this in and then log in” and then “cool you’re all set”. It wrote multiple paragraphs just to communicate “all good”. Not only convoluted but a waste of tokens. I don’t mind the detail when I need it, so I don’t want to change the system instructions. But the judgement of the model to determine the simplicity or directness of response seems low. Writing quality (ex: matching my email tone) still higher with Claude. Ideation and directness or “get to the point”-ness better with GPT-5.6.

Source: Allie K. Miller (X) | 2026-07-29

Loading related sources…