The company Runway announced the launch of Gen-4.5 — a new video generation model using AI. Gen-4.5 is already available to select users and will be open to everyone in the coming days on all pricing plans. The model operates through the Runway platform, offering solutions for both individual creators and organizations worldwide.
Gen-4.5 is distinguished by its high accuracy in motion capture, scene detailing, and ability to adhere to given prompts. It scored 1,247 Elo points on the Artificial Analysis Text to Video leaderboard, surpassing other modern AI models. The generated videos look cinematic and photorealistic, and the model effectively reproduces complex emotions and object interactions.
Users can work with various modes, including “Image to Video,” “Keyframes,” and “Video to Video.” Gen-4.5 allows for detailed camera choreography, complex scenes, and atmosphere changes in a single request. The model maintains the consistency of movements and objects in the frame, which is important for professional video work.
Runway utilizes the power of NVIDIA Hopper and Blackwell to enhance the speed and quality of generation. Thanks to new approaches to training and optimization, Gen-4.5 is suitable for creative tasks, content production, and handling complex video requests.
