Definition
Capability vs efficiency is the distinction between whether a model can solve a problem at all and whether it can solve it cheaply — a model that succeeds with a thousand tokens of reasoning is differently useful from one that succeeds with ten. Conflating the two leads to misleading scaling stories where “new capability” is just “old capability with enough compute.”
Episodes covering this
Worth reading next
Papers we haven't done a deep dive on yet, but would recommend on this topic.