KServe

KServe is a standard model inference platform on Kubernetes, built for highly scalable use cases. It provides performant, standardized inference protocol across ML frameworks including TensorFlow, PyTorch, scikit-learn, XGBoost, and more.

1 APIs 0 Features

InferenceKubernetesMachine LearningMLOpsModel Serving

APIs

KServe Inference API

KServe's standardized model inference protocol for serving predictions across multiple ML frameworks on Kubernetes.

Collections

KServe Inference Protocol

OPEN

Pricing Plans

Kserve Plans Pricing

3 plans

PLANS

Rate Limits

Kserve Rate Limits

5 limits

RATE LIMITS

FinOps

Kserve Finops

FINOPS

Resources

GitHubOrganization

GitHubOrganization

Sources

opencollection: 1.0.0
info:
  name: KServe Inference Protocol
  version: v1+v2
items:
- info:
    name: List models (V1)
    type: http
  http:
    method: GET
    url: http://{host}/v1/models
  docs: List models (V1)
- info:
    name: Model ready (V1)
    type: http
  http:
    method: GET
    url: http://{host}/v1/models/:model_name
    params:
    - name: model_name
      value: ''
      type: path
  docs: Model ready (V1)
- info:
    name: Predict (V1)
    type: http
  http:
    method: POST
    url: http://{host}/v1/models/:model_name:predict
    params:
    - name: model_name
      value: ''
      type: path
    body:
      type: json
      data: '{}'
  docs: Predict (V1)
- info:
    name: Explain (V1)
    type: http
  http:
    method: POST
    url: http://{host}/v1/models/:model_name:explain
    params:
    - name: model_name
      value: ''
      type: path
    body:
      type: json
      data: '{}'
  docs: Explain (V1)
- info:
    name: Server metadata (V2)
    type: http
  http:
    method: GET
    url: http://{host}/v2
  docs: Server metadata (V2)
- info:
    name: Server liveness (V2)
    type: http
  http:
    method: GET
    url: http://{host}/v2/health/live
  docs: Server liveness (V2)
- info:
    name: Server readiness (V2)
    type: http
  http:
    method: GET
    url: http://{host}/v2/health/ready
  docs: Server readiness (V2)
- info:
    name: Model metadata (V2)
    type: http
  http:
    method: GET
    url: http://{host}/v2/models/:model_name
    params:
    - name: model_name
      value: ''
      type: path
  docs: Model metadata (V2)
- info:
    name: Model readiness (V2)
    type: http
  http:
    method: GET
    url: http://{host}/v2/models/:model_name/ready
    params:
    - name: model_name
      value: ''
      type: path
  docs: Model readiness (V2)
- info:
    name: Inference (V2 Open Inference Protocol)
    type: http
  http:
    method: POST
    url: http://{host}/v2/models/:model_name/infer
    params:
    - name: model_name
      value: ''
      type: path
    body:
      type: json
      data: '{}'
  docs: Inference (V2 Open Inference Protocol)
bundled: true