1. This model is called via an API.
2. Each prompt is limited to a maximum of 4 images.
3. Generated images might not always match your expectations.