Exactly. How can "we" develop and encourage benchmarks for multi-turn user assistance? That is what I want. I feel like the models and harnesses push much too hard against this workflow -- that they push you towards letting go and vibe coding, with only your discipline (and desire for a quality and maintainable product) holding it back.

Os there a cost benchmark out there? I wonder how frontier models are doing over time for cost per problem solved.

I think they are optimizing for one-shot performance because that will drive usage. They can’t afford to look bad in the benchmarks. And if that means consuming an order of magnitude more tokens, well, that’s good for business, too.