Runway Image-to-Video API

openapi: 3.0.3
info:
  title: Runway Character Performance Image-to-Video API
  description: 'The Runway (RunwayML) developer API for generative AI media. All generation is asynchronous: a POST to a generation endpoint creates a task and returns a task id, which the client polls on GET /tasks/{id} until the task reaches SUCCEEDED or FAILED. Every request must include an Authorization Bearer API secret and a dated X-Runway-Version header (for example 2024-11-06). Usage is billed in credits (roughly $0.01 per credit) against a developer organization. This description was authored by API Evangelist from Runway''s public documentation and is not an official Runway artifact; verify shapes against docs.dev.runwayml.com.'
  version: '2024-11-06'
  contact:
    name: Runway
    url: https://docs.dev.runwayml.com
servers:
- url: https://api.dev.runwayml.com/v1
  description: Runway developer API
security:
- bearerAuth: []
tags:
- name: Image-to-Video
  description: Generate video from an image and prompt (Gen-4 Turbo, Gen-4.5).
paths:
  /image_to_video:
    post:
      operationId: imageToVideo
      tags:
      - Image-to-Video
      summary: Create an image-to-video task
      description: Generate video from a starting (and optional ending) image and a text prompt. Returns a task id to poll on GET /tasks/{id}.
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/ImageToVideoInput'
      responses:
        '200':
          description: The created task.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/TaskCreated'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '429':
          $ref: '#/components/responses/RateLimited'
components:
  schemas:
    TaskCreated:
      type: object
      properties:
        id:
          type: string
          description: The task id to poll on GET /tasks/{id}.
    ImageToVideoInput:
      type: object
      required:
      - promptImage
      - model
      properties:
        model:
          type: string
          description: Video model id.
          enum:
          - gen4_turbo
          - gen4.5
          - gen3a_turbo
          example: gen4_turbo
        promptImage:
          description: 'A single image URL / data URI, or an array of positioned images ({ uri, position: first|last }) for start/end frames.'
          oneOf:
          - type: string
          - type: array
            items:
              type: object
              properties:
                uri:
                  type: string
                position:
                  type: string
                  enum:
                  - first
                  - last
        promptText:
          type: string
          description: Text prompt guiding the motion / content.
        ratio:
          type: string
          description: Output resolution, for example 1280:768 or 768:1280.
          example: 1280:768
        duration:
          type: integer
          description: Clip length in seconds.
          example: 5
        seed:
          type: integer
          description: Deterministic seed (0 to 4294967295).
  responses:
    RateLimited:
      description: Too many requests - the account rate or concurrency limit was exceeded.
    Unauthorized:
      description: Missing or invalid API secret.
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer
      description: API secret created in the Runway developer portal (dev.runwayml.com), passed as Authorization Bearer YOUR_API_SECRET. Requests must also send a dated X-Runway-Version header.