Modelplane Modelplane docs
Version
main Latest

ModelEndpoint Custom Resource

On this page

A reachable inference endpoint, composed per replica or created manually for external providers.

Concept guide: Route to External Providers →

#Metadata

API version
modelplane.ai/v1alpha1
Kind
ModelEndpoint
Scope
Namespaced
Short names
me

#Example

Manifest
# ModelEndpoints are composed automatically by ModelDeployment.
# Create manually only to register an external inference endpoint.
apiVersion: modelplane.ai/v1alpha1
kind: ModelEndpoint
metadata:
  name: qwen-72b-replica-0
  namespace: ml-team
spec:
  url: http://10.0.1.50/ml-team/qwen-72b/
  rewritePath: /ml-team/qwen-72b/

#Spec

# rewritePath optional string

Path prefix that requests should be rewritten to when routed through this endpoint. Used by ModelService to configure URLRewrite on its HTTPRoute. For Modelplane- composed endpoints this is the per-replica serving path on the remote cluster’s gateway, e.g. /ml-team/qwen-demo/.

# url required string ≥ 1 chars

URL of the inference endpoint. Used to configure routing to this endpoint.

#Status

# conditions optional object[]
# routing optional object

Routing details for this endpoint. ModelService reads backendName to build HTTPRoute backendRefs.

# backendName optional string

Crossplane-generated name of the Backend resource composed by this endpoint.