That's more an issue with prompts and training. For as functional is it is, the established models are all very preliminary, and we're in a Cambrian explosion o

That's more an issue with prompts and training. For as functional is it is, the established models are all very preliminary, and we're in a Cambrian explosion of sorts wrt tooling. Also, one model doesn't speak for them all. I have the problem that my results are often too consistent with certain models, largely because of prompt complexity, lack of wildcarding, etc.