We propose a new architecture called Mixture of Models (MoM) to solve LLM routing for coding workflows. We use a embedding + clustering approach on SWE data and then evaluate LLMs on each cluster to find out who is best.
MoM is a new architecture which has many advantages compared to previous approaches, with MoM you can easily add new released models, no need for re-training,