Runway: Gen-4.5 Text to Video API

Gen-4.5 Text-to-Video is Runway's premier foundation model, representing a paradigm shift in AI cinematography through its Autoregressive-to-Diffusion (A2D) architecture. Unlike its predecessors, Gen-4.5 excels at interpreting complex, multi-layered text prompts involving nuanced human emotions, intricate physical collisions, and precise camera choreography. It introduces a level of "World Intelligence" where objects possess realistic mass, inertia, and causality—such as a character leaning their weight before initiating a run. In 2026, it stands as the highest-performing publicly benchmarked video model, capable of generating cinematic-grade footage that is increasingly indistinguishable from real-world camera captures.

• Input: text
• Output: video
• Released: 2026-02-09
• Knowledge cutoff: 2026-01-31