api-inference/config.json

{
  "framework": "custom",
  "task": "text-to-image",
  "custom_handler": "handler:EndpointHandler"
}