Users say Opus 5 still over-explains simple tasks while GPT-5.6 feels more direct

alliekmiller · x · 2026-07-29

The author says they repeatedly have to tell Opus 5 and Fable 5 to be more direct and less verbose, especially in simple situations like Google login flows.

Main complaints:

the models return multiple paragraphs when a short instruction would do;
they over-explain basic steps, wasting tokens;
the issue is not lack of detail, but poor judgment about when brevity is enough.

The post also gives a subjective comparison:

Claude still feels better for writing quality and matching email tone;
GPT-5.6 feels better for ideation and getting to the point.

Overall, it is a user-level critique of model communication style, not a benchmark or product launch.

Original post →

Users say Opus 5 still over-explains simple tasks while GPT-5.6 feels more direct

More from Models