Runway Unveils Debut World Model and Introduces Native Audio to Video Generator

The competition to launch world models is heating up, with AI video generation leader Runway joining a growing list of tech giants and startups by unveiling its first offering. Called GWM-1, this model operates through frame-by-frame prediction, crafting simulations that comprehend physics and real-world dynamics over time, according to the company.

A world model is an AI that learns an internal representation of how the world functions, enabling it to reason, plan, and act without requiring exhaustive real-world training for every conceivable scenario.

Runway, which recently launched its Gen 4.5 video model that outperformed rivals from Google and OpenAI on the Video Arena leaderboard, claims its GWM-1 world model is more "general-purpose" than Google's Genie-3 and similar competitors. The company positions it as a tool for creating simulations to train AI agents in fields like robotics and life sciences.

"Building a world model first required building an exceptional video model. We believe the most effective path to a world model is teaching models to predict pixels directly, which leads to the best general-purpose simulation. With sufficient scale and the right data, you can develop a model with a deep understanding of how the world operates," explained CTO Anastasis Germanidis during a live stream.

Runway introduced several specialized versions of its new world model: GWM-Worlds, GWM-Robotics, and GWM-Avatars.

GWM-Worlds is an application that lets users create interactive projects. By starting with a text prompt or image reference, users can set a scene; as they explore, the model generates a world with an inherent grasp of geometry, physics, and lighting. Runway notes the simulation runs at 24 fps and 720p resolution. While useful for gaming, Worlds is also designed to teach agents how to navigate and interact within physical environments.

GWM-Robotics aims to leverage synthetic data enhanced with new variables like shifting weather or unexpected obstacles. Runway states this approach can also help identify how and when robots might fail to follow policies or instructions in various situations.

Join the Disrupt 2026 Waitlist

Get on the Disrupt 2026 waitlist for early access to Early Bird tickets. Previous Disrupt stages have featured leaders from Google Cloud, Netflix, Microsoft, Box, Phia, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, and Vinod Khosla — among 250+ experts leading 200+ sessions designed to accelerate your growth and sharpen your competitive edge. You'll also connect with hundreds of startups driving innovation across all industries.

Join the Disrupt 2026 Waitlist

Get on the Disrupt 2026 waitlist for early access to Early Bird tickets. Previous Disrupt stages have featured leaders from Google Cloud, Netflix, Microsoft, Box, Phia, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, and Vinod Khosla — among 250+ experts leading 200+ sessions designed to accelerate your growth and sharpen your competitive edge. You'll also connect with hundreds of startups driving innovation across all industries.

Under GWM-Avatars, Runway is developing realistic human avatars to simulate behavior. This space includes companies like D-ID, Synthesia, Soul Machines, and Google, all working on lifelike avatars for communication, training, and beyond.

The company clarified that Worlds, Robotics, and Avatars are currently separate models, but the long-term plan is to integrate them into a single, unified system.

Alongside the new world model, Runway is enhancing its foundational Gen 4.5 model, released earlier this month. The update adds native audio and long-form, multi-shot generation. Users can now create one-minute videos with consistent characters, native dialogue, background audio, and complex multi-angle shots. The model also allows for editing existing audio, adding dialogue, and modifying multi-shot videos of any length.

This Gen 4.5 update brings Runway closer to competitors like Kling, which also recently launched an all-in-one video suite featuring native audio and multi-shot capabilities. It marks a shift from prototype to production-ready tools for video generation. Runway's updated Gen 4.5 is available to all paid users.

Runway will make GWM-Robotics available via an SDK and is currently in active discussions with multiple robotics companies and enterprises regarding the use of both GWM-Robotics and GWM-Avatars.