name: Runway Generative AI Platform Vocabulary
description: >-
  Core vocabulary, terminology, and domain concepts used in the Runway AI
  developer platform. Covers generative video, image, avatar, and audio APIs
  powered by Runway's advanced foundation models.
created: '2026-05-02'
modified: '2026-05-02'
terms:

  - term: Generative Task
    definition: >-
      An asynchronous job submitted to the Runway API that produces generated media.
      All Runway generation endpoints return a task ID that is polled until status
      is SUCCEEDED, FAILED, or CANCELLED.
    tags:
      - API
      - Async

  - term: Gen-4
    definition: >-
      Runway's fourth-generation video generation model. Supports image-to-video
      generation with high visual quality and controllability. Foundation for
      Gen-4 Turbo and Gen-4.5 variants.
    tags:
      - Model
      - Video Generation

  - term: Gen-4.5
    definition: >-
      An enhanced version of Gen-4 supporting both text-to-video and image-to-video
      generation with improved quality and control. Released February 2026 via API.
      Supports durations from 2-10 seconds.
    tags:
      - Model
      - Video Generation

  - term: Gen-4 Turbo
    definition: >-
      A faster, more cost-efficient variant of Gen-4 with flexible generation
      length from 2-10 seconds. Optimized for production use cases requiring
      rapid generation.
    tags:
      - Model
      - Video Generation

  - term: Gen-4 Aleph
    definition: >-
      A Gen-4 variant supporting video-to-video transformation using an existing
      video as input. Used for video editing and style transfer workflows.
    tags:
      - Model
      - Video Generation

  - term: Veo 3.1
    definition: >-
      Google's video generation model available via the Runway API. Supports
      image-to-video and text-to-video with full 1080p output and first/last
      keyframe support.
    tags:
      - Model
      - Video Generation
      - Google

  - term: Act Two
    definition: >-
      Runway's character performance model that animates characters using reference
      performance video to control facial expressions and body movements. Enables
      motion capture-style animation from video input.
    tags:
      - Model
      - Character Animation

  - term: GWM-1
    definition: >-
      Runway's General World Model, the foundation powering the Characters API.
      Enables real-time conversational avatars with photorealistic or animated
      appearance created from a single reference image.
    tags:
      - Model
      - Characters
      - Avatar

  - term: Avatar
    definition: >-
      A persistent conversational video agent in the Runway Characters API.
      Created from a single reference image with defined appearance, voice, and
      personality. Can reference knowledge documents during conversations.
    tags:
      - Characters
      - Avatar
      - Conversational AI

  - term: Realtime Session
    definition: >-
      A live WebRTC connection between a user and a Runway avatar for real-time
      conversational interaction. Sessions have a maximum duration of 5 minutes.
      WebRTC credentials are single-use.
    tags:
      - Characters
      - WebRTC
      - Real Time

  - term: Text-to-Video
    definition: >-
      A generation workflow that creates a video purely from a text prompt,
      without requiring an input image. Supported by Gen-4.5, Veo 3.1, and
      Veo 3.1 Fast models.
    tags:
      - Video Generation
      - Text to Video

  - term: Image-to-Video
    definition: >-
      A generation workflow that creates a video from an input image, optionally
      guided by a text prompt. The most common Runway generation pattern.
    tags:
      - Video Generation
      - Image to Video

  - term: Video-to-Video
    definition: >-
      A generation workflow that transforms an existing video using a text prompt
      as a transformation guide. Uses the Gen-4 Aleph model.
    tags:
      - Video Generation
      - Video to Video

  - term: Lip Sync
    definition: >-
      A generation workflow that animates a face in an image or video to match
      provided audio. Supports 28+ languages via the ElevenLabs eleven_multilingual_v2
      model integration.
    tags:
      - Video Generation
      - Lip Sync
      - Audio

  - term: Frame Interpolation
    definition: >-
      A video enhancement workflow that increases the frame rate of a video by
      generating intermediate frames, resulting in smoother motion.
    tags:
      - Video Enhancement
      - Frame Interpolation

  - term: Video Upscale
    definition: >-
      A video enhancement workflow that increases the resolution and quality
      of a lower-resolution input video.
    tags:
      - Video Enhancement
      - Upscaling

  - term: Sound Effect Generation
    definition: >-
      A workflow that generates AI-created sound effects for a video or from
      a text description. Optionally synchronized to a provided video.
    tags:
      - Audio
      - Sound Effects

  - term: Bearer Token
    definition: >-
      The authentication method used by all Runway API endpoints. An API key
      is passed in the HTTP Authorization header as "Bearer <api-key>". Obtain
      API keys from the Runway Developer Portal at dev.runwayml.com.
    tags:
      - Authentication
      - Security

  - term: X-Runway-Version
    definition: >-
      A required HTTP header in all Runway API requests specifying the API version.
      Must be set to exactly "2024-11-06". This header enables backward-compatible
      API evolution.
    tags:
      - API
      - Versioning

  - term: Runway Upload
    definition: >-
      A temporary media file uploaded to Runway's storage via a presigned S3 URL.
      Once uploaded, the file is accessible via a runway:// URI that can be used
      as input in generation requests.
    tags:
      - Uploads
      - Media
      - Storage

  - term: Aspect Ratio
    definition: >-
      The proportional relationship between the width and height of generated media.
      Runway supports landscape (1280:720, 1584:672, 1104:832), portrait (720:1280,
      832:1104), and square (960:960) ratios.
    tags:
      - Video Generation
      - Image Generation