Hi Thomas, that's a great question!

Indeed, now every dbt entity (i.e. model, seed, test, or snapshot) will run on their own k8s pod, through KubernetesPodOperator.

However, from my own personal experience, I haven't noticed any significant increase (no more than 5%) in the execution time. The projects I currently maintain consist of around ~1K models each, in case that matters to your specific use-case.