Skip to main content
The Baseten model library provides pre-configured Truss models for deployment to your own account. Unlike Model APIs, which offer instant access to shared endpoints, the model library lets you run dedicated instances on your choice of hardware. Deploying from the model library gives you full control over the infrastructure. You can configure autoscaling, manage cold starts, and deploy to private clusters or specific regions. Use the model library when you need a model that Baseten doesn’t host as an API, or when you want to optimize performance with specific engines and quantization levels.