Leonardo.Ai offers paid users the ability to train their own custom image generation models by uploading a small set of reference images, producing a private model that generates new outputs in a consistent, user-defined style. This feature is available to paid subscribers and is designed for brands, studios, and creators who need style consistency at scale.

Users upload 10–20 reference images representing a specific style, character, or subject. Leonardo.Ai's platform fine-tunes a model on this dataset, capturing nuanced characteristics such as color palette, composition, and aesthetic style. The resulting custom model (built on architectures including Flux Dev / LoRA) can then generate new images consistent with the trained style. Custom models can be kept private or shared within a team (via Leonardo for Teams). This capability is also accessible through the API, where developers can trigger fine-tuning programmatically.