Training a small model on a big one's outputs is producing near-frontier quality at a fraction of the cost — and it has policy makers paying attention too.
Translation for you: the cheap models are catching up fast. If you priced out an AI workflow last year, re-run the math this quarter.