CentML Completions API

openapi: 3.0.1
info:
  title: CentML Chat Completions 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: Completions
paths:
  /completions:
    post:
      operationId: createCompletion
      tags:
      - Completions
      summary: Creates a completion for the provided prompt.
      description: OpenAI-compatible text completions endpoint. Served at https://api.centml.com/openai/v1/completions.
      servers:
      - url: https://api.centml.com/openai/v1
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/CreateCompletionRequest'
      responses:
        '200':
          description: OK
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/CreateCompletionResponse'
components:
  schemas:
    CreateCompletionResponse:
      type: object
      required:
      - id
      - object
      - created
      - model
      - choices
      properties:
        id:
          type: string
        object:
          type: string
          enum:
          - text_completion
        created:
          type: integer
        model:
          type: string
        choices:
          type: array
          items:
            type: object
            properties:
              index:
                type: integer
              text:
                type: string
              finish_reason:
                type: string
                nullable: true
        usage:
          $ref: '#/components/schemas/Usage'
    Usage:
      type: object
      properties:
        prompt_tokens:
          type: integer
        completion_tokens:
          type: integer
        total_tokens:
          type: integer
    CreateCompletionRequest:
      type: object
      required:
      - model
      - prompt
      properties:
        model:
          type: string
        prompt:
          type: string
          description: The prompt to generate completions for.
        max_tokens:
          type: integer
          nullable: true
        temperature:
          type: number
          nullable: true
          default: 1
        top_p:
          type: number
          nullable: true
          default: 1
        n:
          type: integer
          nullable: true
          default: 1
        stream:
          type: boolean
          nullable: true
          default: false
        stop:
          type: array
          nullable: true
          items:
            type: string
  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.'