Runway has unveiled a new artificial intelligence (AI) video generation model, titled Gen-4, as part of its recently launched Gen-4 family. The latest model emphasizes rapid generation while enhancing overall efficiency. According to the AI firm, this innovative model will facilitate quicker iterations and allow users to better explore their creative vision for various projects. The Gen-4 model is now accessible to all subscribers using paid plans. A standout feature of this new video AI model is its improved consistency across characters, settings, and objects within scenes, alongside enhanced adherence to real-world physics.
Runway’s Gen-4 Turbo Released
Runway’s official account announced the debut of the Gen-4 Turbo AI model in a post on X (formerly known as Twitter). The company highlighted that this large language model (LLM) can generate a 10-second video in just 30 seconds, significantly faster compared to the standard Gen-4 model, which may take several minutes for the same duration.
While speed is a key feature of the Gen-4 Turbo, it also offers improved credit efficiency. According to information on Runway’s website, the Turbo model utilizes five credits for each second of video created. Thus, a video lasting five seconds will require 25 credits, while a 10-second clip will use 50 credits. In contrast, the standard Gen-4 model consumes 12 credits per second of video generation.
Runway’s credit system allows users to generate videos based on the subscriptions they select, with different tiers offering varying amounts of credits each month. The most affordable plan, Standard, provides 625 credits for $12 (approximately Rs. 1,034) monthly, while the premium Unlimited tier costs $76 (around Rs. 6,550) per month and grants access to unlimited credits.
The Gen-4 family of video generation models incorporates multiple enhancements compared to its predecessor. Runway claims that the models can create characters that maintain consistency across various lighting conditions, locations, and angles with just one reference image.
Additionally, the Gen-4 model can effectively build scenes utilizing diverse camera angles, including close-ups and wide-angle profiles. It also offers advancements in simulating real-world physics and motion, leading to more realistic representations of elements such as movement, object drops, glass shattering, and wind effects.
