I'm sure those are factors, but it's mainly (1) GPU prices have dropped significantly, and (2) a lot of progress is happening in optimizing large language model

I'm sure those are factors, but it's mainly (1) GPU prices have dropped significantly, and (2) a lot of progress is happening in optimizing large language models to be more efficient while maintaining performance.