While true, I was comparing macro APIs from a usability perspective and should have avoided the term 'abstract model'.
My point was the mental model I refer to when using $lambda is simpler and easier to understand than my mental model for syntax-case, syntax-rules, syntax-parse and friends. The very fact that 'Racket's macros can be entirely compiled away' complicates my mental model which must now accommodate 'phases', compile-time/run-time dichotomies, require for-syntax, etc.
My point was the mental model I refer to when using $lambda is simpler and easier to understand than my mental model for syntax-case, syntax-rules, syntax-parse and friends. The very fact that 'Racket's macros can be entirely compiled away' complicates my mental model which must now accommodate 'phases', compile-time/run-time dichotomies, require for-syntax, etc.