no way, the contents of "reasoning_summary" are summarized?
fyi openai does the same; not really surprising or particularly evil
Not evil but full of hubris
Not evil but full of hubris