GET /health

Check if the server is running. Returns basic status publicly, full diagnostics when authenticated.

Public response (no API key)

curl https://embed.statode.com/health

{
  "status": "healthy"
}

Authenticated response

curl https://embed.statode.com/health -H "X-API-Key: YOUR_KEY"

Field	Type	Description
status	string	Server status ("healthy")
biencoder	string	Bi-encoder model status
backend	string	Inference backend (torch or onnx)
reranker	string or null	Dedup reranker status, null if not loaded
search_reranker	string or null	Search reranker status, null if not loaded
classifier	string or null	Cuisine classifier status, null if not loaded
native_dimension	int	Native embedding dimension of the model
supported_dimensions	int[]	Available embedding dimensions
embedding_cache_size	int	Number of cached embeddings (server-side)
concept_count	int	Number of loaded concept centroids
load_time_seconds	float	Server startup time in seconds

{
  "status": "healthy",
  "biencoder": "loaded",
  "backend": "onnx",
  "reranker": "loaded (cross-encoder)",
  "search_reranker": "loaded",
  "classifier": "loaded",
  "native_dimension": 1024,
  "supported_dimensions": [128, 256, 384],
  "embedding_cache_size": 12847,
  "concept_count": 85,
  "load_time_seconds": 51.3
}

Cost

Free. No credits charged.

Notes

• Use the public endpoint for uptime monitoring and health checks.
• The authenticated response is useful for debugging (confirming which models are loaded, checking cache utilization).
• A 503 response means the server is still loading models at startup. Retry after a few seconds.