Quickstart

Build and open a minimal Model project in Chanterelle in under 5 minutes.

Prerequisites

• Chanterelle installed and configured (Installation guide)
• Python 3.8+

---

Step 1 — Create a project folder

Inside your Chanterelle projects directory, create a new folder:

chanterelle-projects/
  my-first-model/        ← create this

Chanterelle discovers projects by scanning for model_meta.json, analytics.json, or interactive.json files one level deep in the projects directory.

Step 2 — Write model_meta.json

Create my-first-model/model_meta.json:

{
  "model_id": "my-first-model",
  "model_name": "My First Model",
  "model_version": "1.0.0",
  "description_short": "Multiplies two numbers together.",
  "description": "A minimal model that takes two numbers and returns their product.",
  "inputs": [
    {
      "name": "a",
      "label": "Value A",
      "type": "float",
      "required": true,
      "description": "The first number",
      "default": 3.0
    },
    {
      "name": "b",
      "label": "Value B",
      "type": "float",
      "required": true,
      "description": "The second number",
      "default": 4.0
    }
  ],
  "outputs": [
    {
      "name": "product",
      "label": "Product",
      "type": "float",
      "description": "The result of A × B"
    }
  ]
}

Step 3 — Write handler_io.py

Create my-first-model/handler_io.py:

def model_fn():
    """Load your model. The return value is passed to predict_fn as `model`."""
    return None  # No model file needed for this example


def input_fn(inputs: dict, model):
    """Transform raw inputs. The return value is passed to predict_fn as `data`."""
    return inputs


def predict_fn(data: dict, model):
    """Run inference. The return value is passed to output_fn as `prediction`."""
    return {"product": data["a"] * data["b"]}


def output_fn(prediction: dict) -> dict:
    """Format the prediction for display. Return a dict matching your outputs."""
    return prediction

Step 4 — Open in Chanterelle

1. Open Chanterelle.
2. Your project directory is scanned automatically — My First Model should appear in the catalog.
3. Click the project card to open it.
4. You'll see a form with Value A and Value B fields and a Run button.
5. Fill in values and click Run — the Product output appears below.

That's it.

---

What's Next?

Goal	Where to go
Add a dropdown input	category input type
Add preset example inputs	input_presets
Show a chart alongside predictions	model_findings.json
No predictions, just a dashboard	Analytics projects
Conversational agent	Interactive projects
Generate files with AI	AI-Assisted Setup