Use AI Model
Run inference with any public or custom-trained AI model.
The Use AI Model node runs a model on your data and returns predictions. It works with both public pre-built models and models you've trained yourself. Use it for tasks like classification, regression, OCR, object detection, embedding generation, and more.
Configuration
The node uses a cascading selection:
- Category — The AI task (e.g. OCR, Embeddings, Object Detection, Classification)
- Subcategory — A specific task variant within the category
- Model — The model to run
- Model Configuration — Model-specific settings loaded dynamically
Inputs & Outputs
| Name | Description | |
|---|---|---|
| Input | input_data | The data to run inference on (required) |
| Input | trained_model | A custom model from a Train AI Model or Select Artifact node (optional — only needed when using a custom-trained model) |
| Output | predictions | Model predictions as CSV |
| Output | probabilities | Class probabilities for classification tasks |
| Output | confidence_scores | Confidence score per prediction |
Tips
- For custom-trained models, connect a Select Artifact node to the
trained_modelinput - Connect a Preview Output to
predictionsto inspect results directly in the editor - Use upstream Apply Transformation or Custom Code nodes to prepare your input data into the format the model expects