Provider

The Provider node holds one model for image or video generation. Connect it to the Settings input of Generate Image, Generate Video or Generate Video Pro, and that node runs on the Provider's model instead of its own. One Provider node can set the model of many nodes. It never runs a model itself and costs nothing.

Settings

The node shows the name of the model it holds.

Set the model of image or video nodes

On Generate Video and Generate Video Pro, the Settings input also takes Aspect Ratio, Duration and Motion nodes. On Generate Image, it also takes an Aspect Ratio node. See the Settings input of Generate Video and of Generate Image.

Which models a node accepts

Only these three nodes have a Settings input. Set the Provider node's Category to match the nodes it drives: Image for Generate Image, Video for the video nodes.

On a Generate Image node set to several models with Add another model, a connected Provider node replaces the list. The node runs only the Provider node's model, and it is priced for that model only.

When a node cannot run the Provider's model:

• The model's chip on the node turns red.
• Run is disabled, and its tooltip gives the reason, for example "This node can't run Nano Banana. Choose one of its models in the Provider node."
• A run that reaches the node, for example from a workflow run, stops there before anything is charged. The error names the Provider node.

Nodes without a Settings input

Generate Script and the other generation nodes take their model from their own settings. Select the node and choose a model under Provider in the settings panel. On Generate Script and the other nodes that run a language model, the setting is AI Model. The settings below it change to match the model.

To run many shots on one model, drive them from a List into one generation node instead of copying the node. The node runs once per item, so moving from a draft model to a final one is a single change.

Provider nodes in older workflows

A Provider node from an older workflow can hold a category other than Image or Video, or a name that is not a model. When you open its settings, the node moves to a model from the list: the default model of its category, or Seedance 2.0 Fast under Video.

Tips

• Draft on one model, finish on another. Connect one Provider node to every Generate Image node, or to every video node, of a draft. For the final render, change the model once, in the Provider node.
• Compare models side by side. Copy the Provider node, choose another model on the copy, and connect each copy to its own node. To compare image models in one Generate Image node, use Add another model instead, with no Provider node connected.
• Connect one Provider per node. When two Provider nodes are connected to the same Settings input, the one connected last sets the model.
• Choose the model with facts. Choosing a model compares the models for each kind of job, and every model page lists its prices and settings.

Troubleshooting

The model chip is red and Run is disabled. The Provider node holds a model the connected node cannot run. Generate Image refuses a video model, a video node refuses an image model, and Generate Video Pro refuses a video model it does not offer. Set the Provider node's Category to match the node, and choose one of the node's models.

A workflow run stops at a node with an error that names the Provider node. The same cause. The run stopped before anything was charged. Choose a model the node runs, then run again.

A Generate Image node set to several models runs only one. The connected Provider node replaces the node's list of models. Disconnect the Provider node to run every model again.