capabilities.work

Scale an endpoint down to zero replicas

One step with Hugging Face. We can help you get access.

Get something done

Scale an endpoint down to zero replicas

with Hugging Face · Artificial Intelligence Management

Use Hugging Face: its Hugging Face Inference Endpoints API lets you scale an endpoint down to zero replicas in a single step (POST /endpoint/{namespace}/{endpoint_name}/scale-to-zero). capabilities.work can help you get signed up and connected.

  • Can I scale an endpoint to zero replicas so it wakes up again when traffic arrives?
  • What is the difference between scaling to zero and pausing when autoscaling is on?
Say it to your AI assistant
  • Scale endpoint {endpoint_name} in {namespace} to zero replicas.
  • Drop {endpoint_name} under {namespace} to zero and let traffic scale it back up.
For your developer

POST /endpoint/{namespace}/{endpoint_name}/scale-to-zero in Hugging Face Inference Endpoints API. Full spec on APIs.io.

More you can do with Hugging Face

List Inference Endpoints in a namespaceDeploy a new Inference EndpointGet an Inference Endpoint's detailsChange an endpoint's model, hardware or scalingPermanently delete an Inference EndpointPause an endpoint to stop compute costsResume a paused endpointRead an endpoint's runtime logsGet an endpoint's traffic and latency metricsGenerate an image with a chosen Hub modelComplete a prompt via the completions APIGenerate text in one response on TGI