August 23, 2026
Before Fable, spending too much time improving your programming toolchain or context strategies always felt a little absurd. A new model would come along at the same price (or even cheaper!) and paper over most of your problems.
But then Fable was released. It was—and still is!—incredible. But its cost was simply too high, while Opus was already “good enough” (as were 5.6, K3, and even GLM) to handle most of the coding work we needed.
So we started thinking about what should handle different kinds of work.
— Drew Breunig, “Fable and the End of the Free Lunch”