MMOOZZGG.com

Glossary

Runway AI

In short

Runway is a platform from Runway AI, Inc. for creating and editing video with generative models. Its video models include Gen-4.5.

Models and products · Runway

MMOOZZGG.com editorsChecked: 1 min read

For a child

Imagine a box of magic crayons: you draw over a photo with them, and the picture comes alive, the clouds float and the dog wags its tail. You can even draw a whole little world and walk around in it. But the crayons sometimes get confused: an object suddenly disappears and then appears again. So people always look closely at what came out.

In plain words

It is an online studio where video is created and edited with AI. You describe a scene in text, and the model generates a clip; the company promised to bring the image, keyframe and source video modes to the new model as well. Besides generation, Runway develops world models: they build an interactive environment frame by frame, and you can move through it in real time. For example, an editor describes a camera flight over a city at night and gets a background for the credits instead of buying a stock clip. The service works in the browser and in mobile apps.

For business

Agencies and production studios use such services for previsualization, backgrounds and draft scenes before an expensive shoot. Under the terms of use, Runway claims no rights to your inputs and outputs and does not limit their commercial use, but it gets a broad license to them, including for training its models. This matters for projects with client materials. The company publishes no list of supported countries; the terms updated on 15 September 2026 forbid exporting the service to countries under US embargo and providing it to persons on sanctions lists. The weights of the current models are not open. How employees should set tasks for such generators is covered in generative AI training.

For a specialist

Runway is a family of generative video models and services. Gen-4.5 came out on 1 December 2025: the company notes gains in physical accuracy, motion quality and temporal consistency, and lists known limits such as broken causality, failures of object permanence and actions in the frame that succeed too often. The model was developed on NVIDIA GPUs, and inference runs on Hopper and Blackwell. On 11 December 2025 the company released GWM-1, its first family of general world models built on Gen-4.5: the Worlds, Avatars and Robotics variants generate video in real time in response to actions. The early scientific basis is described in the paper by Esser et al., “Structure and Content-Guided Video Synthesis with Diffusion Models” (2023): a diffusion model edits existing video from text or an image and keeps the structure of the scene through depth maps.

Example

A studio is shooting a furniture commercial and, before the shoot, builds draft scenes with different lighting so the client can choose the mood. The team arrives on set with an approved storyboard.

Related terms

Questions and answers

What can Runway do? The platform creates video from a text description and edits clips with AI. The company also develops world models that build interactive scenes in real time.
Who owns the videos created in Runway? Under the terms of use, the company claims no rights to your inputs and outputs and does not limit their commercial use. At the same time, the user grants Runway a broad license to these materials, including for model training.

Sources

Not sure whether your company needs this?

AI solutions map
Contact us