CentML Deployments API

openapi: 3.0.1
info:
  title: CentML Chat Deployments API
  description: 'Specification of the CentML API surface. CentML exposes two related but distinct HTTP APIs: (1) an OpenAI-compatible serverless inference API for chat completions, text completions, and model listing served at https://api.centml.com/openai/v1, authenticated with a Bearer serverless token; and (2) a platform (control-plane) API for managing dedicated inference / compute / job deployments and clusters, also Bearer-authenticated. Both are modeled here. Endpoints reflect CentML''s published documentation (docs.centml.ai) and the official platform_api_python_client.'
  termsOfService: https://centml.ai/terms-of-service/
  contact:
    name: CentML Support
    url: https://docs.centml.ai/resources/requesting_support
  version: '1.0'
servers:
- url: https://api.centml.com/openai/v1
  description: OpenAI-compatible serverless inference base URL.
- url: https://api.centml.com
  description: CentML platform (control-plane) base URL for deployments and clusters.
security:
- bearerAuth: []
tags:
- name: Deployments
paths:
  /deployments:
    get:
      operationId: listDeployments
      tags:
      - Deployments
      summary: Lists all deployments for the account.
      description: Platform control-plane endpoint. Returns all deployments (inference, compute, and job) for the authenticated account. Served at https://api.centml.com/deployments.
      servers:
      - url: https://api.centml.com
      parameters:
      - name: type
        in: query
        required: false
        description: Optional deployment type filter (e.g. inference, compute, job).
        schema:
          type: string
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/DeploymentList'
  /deployments/inference:
    post:
      operationId: createInferenceDeployment
      tags:
      - Deployments
      summary: Creates a dedicated inference deployment.
      description: Platform control-plane endpoint. Creates a dedicated, autoscaling inference deployment that serves a model as an HTTPS endpoint on a selected cluster and hardware instance. Served at https://api.centml.com/deployments/inference.
      servers:
      - url: https://api.centml.com
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/CreateInferenceDeploymentRequest'
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Deployment'
    put:
      operationId: updateInferenceDeployment
      tags:
      - Deployments
      summary: Updates an existing inference deployment.
      servers:
      - url: https://api.centml.com
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/CreateInferenceDeploymentRequest'
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Deployment'
  /deployments/inference/{deployment_id}:
    get:
      operationId: getInferenceDeployment
      tags:
      - Deployments
      summary: Retrieves an inference deployment by ID.
      servers:
      - url: https://api.centml.com
      parameters:
      - name: deployment_id
        in: path
        required: true
        description: The ID of the deployment to retrieve.
        schema:
          type: integer
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Deployment'
  /deployments/compute:
    post:
      operationId: createComputeDeployment
      tags:
      - Deployments
      summary: Creates a compute deployment.
      description: Platform control-plane endpoint to create a compute deployment. Served at https://api.centml.com/deployments/compute.
      servers:
      - url: https://api.centml.com
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/CreateComputeDeploymentRequest'
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Deployment'
  /deployments/status/{deployment_id}:
    get:
      operationId: getDeploymentStatus
      tags:
      - Deployments
      summary: Gets the status of a deployment.
      servers:
      - url: https://api.centml.com
      parameters:
      - name: deployment_id
        in: path
        required: true
        schema:
          type: integer
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/DeploymentStatus'
    put:
      operationId: updateDeploymentStatus
      tags:
      - Deployments
      summary: Updates the status of a deployment (e.g. pause or delete).
      servers:
      - url: https://api.centml.com
      parameters:
      - name: deployment_id
        in: path
        required: true
        schema:
          type: integer
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/DeploymentStatus'
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/DeploymentStatus'
  /deployments/logs/{deployment_id}:
    get:
      operationId: getDeploymentLogs
      tags:
      - Deployments
      summary: Returns logs for a deployment.
      servers:
      - url: https://api.centml.com
      parameters:
      - name: deployment_id
        in: path
        required: true
        schema:
          type: integer
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/DeploymentLogs'
components:
  schemas:
    DeploymentList:
      type: object
      properties:
        results:
          type: array
          items:
            $ref: '#/components/schemas/Deployment'
    DeploymentStatus:
      type: object
      properties:
        id:
          type: integer
        status:
          type: string
          description: The deployment status to read or set (e.g. running, paused, deleted).
    CreateInferenceDeploymentRequest:
      type: object
      required:
      - name
      - cluster_id
      - hardware_instance_id
      properties:
        name:
          type: string
          description: A name for the inference deployment.
        cluster_id:
          type: integer
          description: ID of the cluster to deploy to (from GET /clusters).
        hardware_instance_id:
          type: integer
          description: ID of the GPU hardware instance (from GET /capacity).
        model:
          type: string
          description: The model or image to serve.
        min_replicas:
          type: integer
          description: Minimum number of replicas for autoscaling.
        max_replicas:
          type: integer
          description: Maximum number of replicas for autoscaling.
    Deployment:
      type: object
      properties:
        id:
          type: integer
          description: The unique deployment ID.
        name:
          type: string
          description: Human-readable deployment name.
        type:
          type: string
          description: Deployment type (inference, compute, or job).
        status:
          type: string
          description: Current deployment status (e.g. running, paused, deleted).
        cluster_id:
          type: integer
          description: The cluster the deployment is placed on.
        endpoint_url:
          type: string
          description: The HTTPS endpoint URL once the deployment is ready.
    CreateComputeDeploymentRequest:
      type: object
      required:
      - name
      - cluster_id
      - hardware_instance_id
      properties:
        name:
          type: string
        cluster_id:
          type: integer
        hardware_instance_id:
          type: integer
    DeploymentLogs:
      type: object
      properties:
        logs:
          type: string
          description: Log output for the deployment.
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer
      bearerFormat: CentML API key
      description: 'Set the `Authorization: Bearer <CENTML_API_KEY>` header. Serverless inference uses a serverless token; the platform API uses an account access token. Both are issued from the CentML platform.'