Actions16
- LLM (Text Generation) Actions
- Image Generation Actions
- Video Generation Actions
- Text-to-Speech Actions
- Speech-to-Text Actions
- Music Generation Actions
- Embeddings Actions
- Content Moderation Actions
- Custom Inference Actions
Custom Inference → Predict
AI-generatedSummary
Send input data to a specified deployed model on Chutes.ai for custom inference to obtain prediction results.
Inputs
- Chute URL — Select a specific chute URL to route the inference request; defaults to https://llm.chutes.ai if not specified.
- Model ID (required) — The unique identifier of the deployed model on Chutes.ai to use for prediction.
- Input (required) — Input data provided in JSON format to be sent to the model for inference.
- Additional Options — Optional parameters including Maximum Timeout in seconds (to limit inference wait time), Webhook URL for asynchronous result delivery, priority of the inference job (Low, Normal, High), and desired output format (JSON or Raw).
Output shape
a single JSON object containing the prediction results from the inference model
The output format can be controlled via the 'Output Format' option, and the operation routes requests to the specified chute and model. If a timeout is set and exceeded, the node will error and can trigger a retry workflow. Asynchronous inference can be used by specifying a webhook URL.
Examples
Example 1: Invoke a text classification model with JSON input
Set Model ID to 'model_abc123', provide JSON input like {"text": "Hello world"}, optionally select a Chute URL or use default, and adjust additional options if needed.