HTTP API Reference

Default base URL: http://localhost:11434

POST /api/predict

Run inference on a loaded model.

Request

{
  "model": "my-model",
  "inputs": [[1.0, 2.0, 3.0, ...]]
}

Field	Type	Required	Description
model	string	Yes	Name of the loaded model
inputs	float[][]	Yes	2D array, shape [n_samples, n_features]

Response (200 OK)

{
  "model": "my-model",
  "outputs": [0.97],
  "n_samples": 1,
  "latency_us": 91.0,
  "done": true
}

Field	Type	Description
model	string	Model name
outputs	float[]	Prediction values, length = n_samples × n_outputs
n_samples	int	Number of samples processed
latency_us	float	Inference latency in microseconds (C call only)
done	bool	Always true for non-streaming responses

Errors

{"error": "model 'xyz' not loaded"}          // 404
{"error": "expected 30 features, got 10"}     // 400
{"error": "missing 'inputs' field"}           // 400
{"error": "missing 'model' field"}            // 400

Batch Example

Send multiple samples in one request:

curl http://localhost:11434/api/predict \
  -d '{
    "model": "my-model",
    "inputs": [
      [1.0, 2.0, 3.0],
      [4.0, 5.0, 6.0],
      [7.0, 8.0, 9.0]
    ]
  }'

---

POST /api/generate

Alias for /api/predict. Provided for compatibility with Ollama client libraries.

---

GET /api/models

List all loaded models with metadata.

Response (200 OK)

{
  "models": [
    {
      "name": "fraud-detector",
      "n_features": 30,
      "n_outputs": 1,
      "n_trees": 50,
      "objective": "binary:logistic",
      "framework": "xgboost",
      "format": "xgboost",
      "version": "0.1.0"
    }
  ]
}

---

GET /api/model/:name

Get metadata for a specific model.

Response (200 OK)

{
  "name": "fraud-detector",
  "n_features": 30,
  "n_outputs": 1,
  "n_trees": 50,
  "objective": "binary:logistic",
  "framework": "xgboost"
}

Error (404)

{"error": "model 'xyz' not found"}

---

GET /api/health

Health check endpoint.

Response (200 OK)

{"status": "ok", "version": "0.1.0"}