I feel that they've added a classifier model, where it will determine whether a satisfactory answer to a prompt will require the full model or a smaller one. Th

I feel that they've added a classifier model, where it will determine whether a satisfactory answer to a prompt will require the full model or a smaller one. This leads to the inconsistent quality of outputs: some are really from a worse model that the system believes it can get away with and still be "GPT-4" quality. When it takes a long time to get an answer from GPT-4, it seems better, but worse when it's unexpectedly fast.