Interesting concept, but I'm skeptical about the value of aggregating models vs. fine-tuning a few good ones. Have you considered offering fine-tuning services instead?
so thats exactly our plan going forward-once we see usage patterns of our user's, we will fine-tune models and also offer intelligent router to further optimize for speed and performance for the tasks.
