> The cuts mark a shift for US AI groups that make proprietary “closed” models that have, until now, competed heavily on performance.
They've been competing on price and performance since the early days. I remember when the GPT-3 models were SotA (when journalists were still writing about them in terms of "perhaps one day they'll be able to write this very article" rather than "this is an economic threat") and there being a range of models (ada/babbage/curie/davinci) with different price/quality tradeoffs.
Ever since, it's been a mix of bleeding-edge performance at high prices and distilled models at huge discounts if you can wait a few months.
> The cuts mark a shift for US AI groups that make proprietary “closed” models that have, until now, competed heavily on performance.
They've been competing on price and performance since the early days. I remember when the GPT-3 models were SotA (when journalists were still writing about them in terms of "perhaps one day they'll be able to write this very article" rather than "this is an economic threat") and there being a range of models (ada/babbage/curie/davinci) with different price/quality tradeoffs.
Ever since, it's been a mix of bleeding-edge performance at high prices and distilled models at huge discounts if you can wait a few months.