Did you use the LLM aggregator like https://model.box ?

Cool to see a practical application of model routing working under the hood - wonder how it works with different prompts

This is big. I was using notdiamond in a personal project but this is such a great use case for it