Huh. I find myself preferring aspects of TEXT-DAVINCI-001 in just about every example prompt.

It’s so to-the-point. No hype. No overstepping. Sure, it lacks many details that later models add. But those added details are only sometimes helpful. Most of the time, they detract.

Makes me wonder what folks would say if you re-released TEXT-DAVINCI-001 as “GPT5-BREVITY”

I bet you’d get split opinions on these types of not so hard / creative prompts.