Which statement is NOT true on ensemble models?

Empirically, ensemble models tend to have better performance than standalone machine learning models.

Ensemble models are when more than one model is used to predict an outcome.

Examples of ensemble models can include random forest and gradient boosting.

Ensemble models are always a combination of multiple decision trees.