Parasail Deployments API

openapi: 3.1.0
info:
  title: Parasail Batch Deployments API
  description: 'OpenAI-compatible Batch API for offline, asynchronous inference workloads at 50% off

    serverless pricing (with cached tokens at an additional 30% off). Supports

    /v1/chat/completions and /v1/embeddings in the OpenAI Batch file format (JSONL).

    '
  version: '1.0'
  contact:
    name: Parasail
    url: https://docs.parasail.io/parasail-docs/
servers:
- url: https://api.parasail.io/v1
  description: Parasail Batch endpoint
security:
- bearerAuth: []
tags:
- name: Deployments
  description: Manage dedicated GPU deployments for custom and reserved-capacity inference.
paths:
  /dedicated/deployments:
    get:
      tags:
      - Deployments
      operationId: listDeployments
      summary: List Deployments
      description: List all dedicated deployments owned by the authenticated organization.
      responses:
        '200':
          description: A list of dedicated deployments.
          content:
            application/json:
              schema:
                type: array
                items:
                  $ref: '#/components/schemas/Deployment'
    post:
      tags:
      - Deployments
      operationId: createDeployment
      summary: Create Deployment
      description: Provision a new dedicated deployment of a chosen model on a chosen GPU device configuration.
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/DeploymentCreateRequest'
      responses:
        '201':
          description: The created deployment.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Deployment'
  /dedicated/deployments/{deployment_id}:
    get:
      tags:
      - Deployments
      operationId: getDeployment
      summary: Retrieve Deployment
      description: Retrieve a dedicated deployment by ID.
      parameters:
      - in: path
        name: deployment_id
        required: true
        schema:
          type: string
      responses:
        '200':
          description: The deployment.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Deployment'
    put:
      tags:
      - Deployments
      operationId: updateDeployment
      summary: Update Deployment
      description: Update mutable properties of a dedicated deployment (e.g. replicas, autoscaling).
      parameters:
      - in: path
        name: deployment_id
        required: true
        schema:
          type: string
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/DeploymentUpdateRequest'
      responses:
        '200':
          description: The updated deployment.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Deployment'
    delete:
      tags:
      - Deployments
      operationId: deleteDeployment
      summary: Delete Deployment
      description: Permanently delete a dedicated deployment and release its reserved GPU capacity.
      parameters:
      - in: path
        name: deployment_id
        required: true
        schema:
          type: string
      responses:
        '204':
          description: Deployment deleted.
  /dedicated/deployments/{deployment_id}/pause:
    post:
      tags:
      - Deployments
      operationId: pauseDeployment
      summary: Pause Deployment
      description: Pause a dedicated deployment to stop billing while preserving configuration.
      parameters:
      - in: path
        name: deployment_id
        required: true
        schema:
          type: string
      responses:
        '200':
          description: The paused deployment.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Deployment'
  /dedicated/deployments/{deployment_id}/resume:
    post:
      tags:
      - Deployments
      operationId: resumeDeployment
      summary: Resume Deployment
      description: Resume a previously-paused dedicated deployment.
      parameters:
      - in: path
        name: deployment_id
        required: true
        schema:
          type: string
      responses:
        '200':
          description: The resumed deployment.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Deployment'
components:
  schemas:
    DeploymentUpdateRequest:
      type: object
      properties:
        replicas:
          type: integer
        autoscaling:
          type: object
          properties:
            min_replicas:
              type: integer
            max_replicas:
              type: integer
    DeploymentCreateRequest:
      type: object
      required:
      - deploymentName
      - modelName
      - deviceConfigs
      - replicas
      properties:
        deploymentName:
          type: string
          description: Human-readable name for the deployment.
        modelName:
          type: string
          description: Hugging Face model name or Parasail-recognised model ID to serve.
        deviceConfigs:
          type: array
          description: GPU device configurations to provision (count and SKU).
          items:
            type: object
            properties:
              device:
                type: string
                description: GPU SKU identifier (e.g. H100, A100, H200).
              count:
                type: integer
        replicas:
          type: integer
          description: Number of replicas to run.
        autoscaling:
          type: object
          properties:
            min_replicas:
              type: integer
            max_replicas:
              type: integer
    Deployment:
      type: object
      properties:
        deployment_id:
          type: string
        deploymentName:
          type: string
        modelName:
          type: string
        status:
          type: string
          enum:
          - provisioning
          - running
          - paused
          - updating
          - deleting
          - failed
        deviceConfigs:
          type: array
          items:
            type: object
            properties:
              device:
                type: string
              count:
                type: integer
        replicas:
          type: integer
        endpoint_url:
          type: string
          description: The OpenAI-compatible inference URL for this dedicated deployment.
        created_at:
          type: string
          format: date-time
        updated_at:
          type: string
          format: date-time
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer
      description: Send your Parasail API key as a Bearer token.