Runway conquers the video generation landscape with consistency thanks to its Gen-4 model

Runway Research has just unveiled Gen-4, the latest evolution of their image and video generating AI. This new version stands out for its ability to maintain consistency in the pieces it creates, enabling the generation of video sequences where characters and environments retain a stable design and style, even when there are changes in the scene.

In addition to achieving consistency in subjects, objects, and style, Gen-4 represents a qualitative leap compared to its predecessor Gen-3 Alpha, as it possesses an enhanced understanding of language and allows the generation of videos with a higher level of dynamism featuring realistic motion.

The Image-to-Video feature of Gen-4 is already being activated for all paid plans and enterprise clients.

Today we are introducing Gen-4, our new series of state-of-the-art AI models for media generation and world consistency. Gen-4 is a significant step forward for fidelity, dynamic motion, and controllability in generative media.

Gen-4 Image-to-Video is rolling out today to all paid… pic.twitter.com/VKnY5pWC8X

— Runway (@runwayml) March 31, 2025

Strengths of Runway Gen-4

Consistent Characters and Objects

The Gen-4 model offers the capability of infinite character consistency from a single reference image. Upload an image of a person or character to the tool and start creating a world of narratives where this subject maintains its attributes regardless of lighting conditions, location, and frame treatment.

The same applies to objects. Place any element in the desired location or condition, and it will maintain its original appearance. This can be particularly useful when generating promotional videos for products.

Coverage

The AI of Gen-4 is capable of capturing all angles of any scene. Simply attach a reference image and describe, using text, the composition of the shot you wish to generate.

Superior Quality and Understanding

Gen-4’s understanding capabilities surpass those of its predecessors, enhancing its ability to meet user requests when generating videos. Additionally, the model also presents significant advancements in the quality of the works it creates.

Understanding of Real-World Physics

Runway asserts that Gen-4 represents a further step towards “Universal Generative Models” (versatile AI models capable of generating various types of content and not focused or trained for a single task) capable of understanding the world. Indeed, this new video-generating AI reveals a greater potential to simulate real-world physics in its creations.

Visual Effects

With this tool, it is possible to generate videos quickly, controlled, and flexibly, integrating action content and visual effects.

Runway has published a series of short films entirely created with Gen-4 to demonstrate the capabilities of their new model. Here is one about a charming little flame that will warm your heart.

Photo: Runway