I've also seen the opposite, where tiny little 7B models get real close to GPT4 quality results on really specifically use cases. If you're trying to scale jus

I've also seen the opposite, where tiny little 7B models get real close to GPT4 quality results on really specifically use cases. If you're trying to scale just that use case it's significantly cheaper, and also faster to just scale up inference with that specialty model. An example of this is using an LLM to extract medical details from a record.