Runway Unveils Real-Time AI Video Generation with GWM-1 and Gen-4.5

Runway has announced its GWM-1, the first General World Model, alongside the Gen-4.5 video generation model, enabling real-time video generation. This development follows the introduction of GWM-1 in December 2025 and the initial discussions on real-time capabilities in March 2026. The Gen-4.5 model aims to deliver the first frame in under 100 milliseconds, significantly improving the responsiveness of video generation. Runway claims that real-time generation could reduce waiting times and GPU costs, effectively closing the gap between conceptualization and execution of video content.

Runway's approach leverages partnerships with Nvidia for real-time model development, enhancing the capabilities of their Solaris system for generating user interfaces. The company emphasizes that interactive applications represent the most significant long-term use case for AI-generated media. In a related context, Waymo has reported that its Driver has traveled billions of miles in virtual worlds, utilizing the Waymo World Model for simulations, which underscores the growing importance of real-time models in AI applications. This shift towards real-time capabilities is expected to transform how practitioners build and deploy AI-driven video content.