The dataset contains publicly licensed free models from asset stores, as well as models purchased through various legitimate channels. Importantly, we do not scrape or utilize privately owned models for training purposes.