OpenRelay Clusters API

openapi: 3.1.0
info:
  description: 'The OpenRelay control-plane REST API. Deploy GPU VMs and inference clusters, manage organizations and billing, and automate your infrastructure. All requests are authenticated with an API key (`vl_…`) unless noted otherwise.

    '
  title: OpenRelay Account Clusters API
  version: 0.1.0
servers:
- description: Production
  url: https://api.openrelay.inc
- description: Beta
  url: https://api.beta.openrelay.inc
- description: Local development
  url: http://localhost:8083
tags:
- description: Autoscaling inference clusters that serve a container image behind an endpoint.
  name: Clusters
paths:
  /v1/clusters/{id}:
    get:
      operationId: getCluster
      parameters:
      - in: path
        name: id
        required: true
        schema:
          type: string
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ClusterSummary'
          description: The cluster
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/Forbidden'
        '404':
          $ref: '#/components/responses/NotFound'
      security:
      - apiKey: []
      summary: Get a cluster by id
      tags:
      - Clusters
  /v1/clusters/{id}/detail:
    get:
      operationId: getClusterDetail
      parameters:
      - in: path
        name: id
        required: true
        schema:
          type: string
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ClusterDetail'
          description: The cluster detail
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/Forbidden'
        '404':
          $ref: '#/components/responses/NotFound'
      security:
      - apiKey: []
      summary: Get a cluster with its replicas and GPU model
      tags:
      - Clusters
      x-openrelay-mcp:
        default: true
  /v1/clusters/{id}/restart:
    post:
      operationId: restartCluster
      parameters:
      - in: path
        name: id
        required: true
        schema:
          type: string
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ClusterSummary'
          description: OK
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/Forbidden'
        '404':
          $ref: '#/components/responses/NotFound'
      security:
      - apiKey: []
      summary: Restart a cluster
      tags:
      - Clusters
  /v1/clusters/{id}/scale:
    post:
      operationId: scaleCluster
      parameters:
      - in: path
        name: id
        required: true
        schema:
          type: string
      requestBody:
        content:
          application/json:
            example:
              replicaCount: 4
            schema:
              $ref: '#/components/schemas/ScaleRequest'
        required: true
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ClusterSummary'
          description: OK
        '400':
          $ref: '#/components/responses/BadRequest'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/Forbidden'
        '404':
          $ref: '#/components/responses/NotFound'
        '409':
          $ref: '#/components/responses/Conflict'
      security:
      - apiKey: []
      summary: Scale a cluster's replica count
      tags:
      - Clusters
  /v1/clusters/{id}/stop:
    post:
      operationId: stopCluster
      parameters:
      - in: path
        name: id
        required: true
        schema:
          type: string
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ClusterSummary'
          description: OK
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/Forbidden'
        '404':
          $ref: '#/components/responses/NotFound'
      security:
      - apiKey: []
      summary: Stop a cluster
      tags:
      - Clusters
  /v1/clusters/{id}/terminate:
    post:
      operationId: terminateCluster
      parameters:
      - in: path
        name: id
        required: true
        schema:
          type: string
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ClusterSummary'
          description: OK
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/Forbidden'
        '404':
          $ref: '#/components/responses/NotFound'
      security:
      - apiKey: []
      summary: Terminate a cluster
      tags:
      - Clusters
      x-openrelay-mcp:
        destructiveHint: true
  /v1/orgs/{orgId}/clusters:
    get:
      operationId: listOrgClusters
      parameters:
      - in: path
        name: orgId
        required: true
        schema:
          type: string
      - in: query
        name: limit
        required: false
        schema:
          type: integer
      - in: query
        name: cursor
        required: false
        schema:
          type: string
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ClusterPage'
          description: Page of clusters
        '400':
          $ref: '#/components/responses/BadRequest'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/Forbidden'
      security:
      - apiKey: []
      summary: List an org's clusters (cursor-paginated)
      tags:
      - Clusters
      x-openrelay-mcp:
        default: true
  /v1/orgs/{orgId}/clusters/create:
    post:
      operationId: createCluster
      parameters:
      - in: path
        name: orgId
        required: true
        schema:
          type: string
      requestBody:
        content:
          application/json:
            example:
              containerPort: 8000
              gpuModelId: rtx-4090
              gpusPerReplica: 1
              healthCheckPath: /health
              imageUrl: vllm/vllm-openai:latest
              name: vllm-serving
              replicaCount: 2
            schema:
              $ref: '#/components/schemas/CreateClusterRequest'
        required: true
      responses:
        '200':
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/ClusterSummary'
          description: Created
        '400':
          $ref: '#/components/responses/BadRequest'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/Forbidden'
        '409':
          $ref: '#/components/responses/Conflict'
      security:
      - apiKey: []
      summary: Create a cluster
      tags:
      - Clusters
      x-openrelay-mcp:
        default: true
components:
  schemas:
    ClusterDetail:
      properties:
        allowFallback:
          type: boolean
        containerPort:
          type: integer
        createdAt:
          type: string
        deployToken:
          type: string
        diskSizeGb:
          type: integer
        endpointUrl:
          type: string
        envVars:
          additionalProperties:
            type: string
          type: object
        gpuModel:
          $ref: '#/components/schemas/GpuModel'
        gpuModelId:
          type: string
        gpusPerReplica:
          type: integer
        healthCheckPath:
          type: string
        healthCheckPort:
          type: integer
        healthChecks:
          items:
            type: string
          type: array
        id:
          type: string
        imageUrl:
          type: string
        name:
          type: string
        notes:
          type: string
        organizationId:
          type: string
        preferredRegions:
          items:
            type: string
          type: array
        privileged:
          type: boolean
        replicaCount:
          type: integer
        replicas:
          items:
            $ref: '#/components/schemas/ClusterReplicaItem'
          type: array
        resourceSize:
          type: string
        status:
          type: string
        statusReason:
          description: human-readable cause for failed/stuck states
          type: string
        tier:
          type: string
        updatedAt:
          type: string
      required:
      - id
      - organizationId
      - name
      - status
      - gpusPerReplica
      - resourceSize
      - replicaCount
      - tier
      - allowFallback
      - diskSizeGb
      - privileged
      - replicas
      - createdAt
      - updatedAt
      type: object
    GpuModel:
      properties:
        architecture:
          type: string
        createdAt:
          type: string
        hidden:
          type: boolean
        id:
          type: string
        manufacturer:
          type: string
        name:
          type: string
        vramGb:
          type: integer
      required:
      - id
      - name
      - manufacturer
      - vramGb
      - hidden
      type: object
    ClusterSummary:
      properties:
        activeReplicas:
          type: integer
        createdAt:
          type: string
        endpointUrl:
          type: string
        gpuModelId:
          type: string
        id:
          type: string
        name:
          type: string
        organizationId:
          type: string
        replicaCount:
          type: integer
        status:
          type: string
      required:
      - id
      - organizationId
      - name
      - status
      type: object
    ClusterReplicaItem:
      properties:
        consecutiveFailures:
          type: integer
        createdAt:
          type: string
        diskTotalBytes:
          format: int64
          type: integer
        diskUpdatedAt:
          type: string
        diskUsedBytes:
          format: int64
          type: integer
        endpointUrl:
          type: string
        gpuCount:
          type: integer
        healthy:
          type: boolean
        id:
          type: string
        internalIp:
          type: string
        lastHealthCheckAt:
          type: string
        nodeHostname:
          type: string
        nodeId:
          type: string
        nodeLocation:
          type: string
        nodeRegion:
          type: string
        nvlinkEnabled:
          type: boolean
        port:
          type: integer
        sshPort:
          type: integer
        startedAt:
          type: string
        status:
          type: string
      required:
      - id
      - gpuCount
      - status
      - healthy
      - nvlinkEnabled
      - createdAt
      type: object
    ScaleRequest:
      properties:
        replicaCount:
          minimum: 1
          type: integer
      required:
      - replicaCount
      type: object
    ClusterPage:
      properties:
        items:
          items:
            $ref: '#/components/schemas/ClusterSummary'
          type: array
        nextCursor:
          type: string
      required:
      - items
      type: object
    Error:
      properties:
        code:
          type: string
        error:
          type: string
      required:
      - error
      type: object
    CreateClusterRequest:
      properties:
        allowFallback:
          type: boolean
        containerCommand:
          items:
            type: string
          type: array
        containerPort:
          type: integer
        diskSizeGb:
          type: integer
        envVars:
          additionalProperties:
            type: string
          type: object
        gpuModelId:
          type: string
        gpusPerReplica:
          type: integer
        healthCheckPath:
          type: string
        healthCheckPort:
          type: integer
        healthChecks:
          items:
            type: string
          type: array
        imageUrl:
          type: string
        kernelModules:
          items:
            type: string
          type: array
        name:
          type: string
        preferredRegions:
          items:
            type: string
          type: array
        privileged:
          type: boolean
        registryCredentialId:
          type: string
        replicaCount:
          type: integer
        resourceSize:
          type: string
        templateId:
          type: string
        tier:
          type: string
      required:
      - name
      type: object
  responses:
    Forbidden:
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/Error'
      description: API key lacks the required scope
    NotFound:
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/Error'
      description: Resource not found
    Conflict:
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/Error'
      description: The request conflicts with existing state
    Unauthorized:
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/Error'
      description: Missing or invalid API key
    BadRequest:
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/Error'
      description: The request is invalid
  securitySchemes:
    apiKey:
      description: 'OpenRelay API key. Send it as `Authorization: Bearer vl_…`.'
      scheme: bearer
      type: http