Runway Introduces Gen-4 Turbo Video AI Model With Faster Generation Speeds

Runway on Monday introduced a new artificial intelligence (AI) video generation model in the recently released Gen-4 family. Dubbed Gen-4 Turbo, the latest AI model prioritises generation speed while optimising efficiency. The AI firm says this AI model will help users in faster iterations and exploring creative vision for a project. The new addition to the Gen-4 family is now available across all paid plans. Notably, the new generation of video AI models offers improved consistency across characters, locations, and objects throughout the scene, as well as better real-world physics.

Runway's Gen-4 Turbo Comes Released

In a post on X (formerly Twitter), Runway said that the latest Gen-4 Turbo large language model (LLM) takes 30 seconds to generate a 10-second-long video. In comparison, the standard Gen-4 model can take up to a couple of minutes to generate a video of the same duration.

While improved speed is the main highlight of the AI model, it is also more credit-efficient. According to Runway's website, the Gen-4 Turbo consumes five credits per second of video generation. A five-second-long video will consume 25 credits, and a 10-second-long video will consume 50 credits. In comparison, the Gen-4 AI model consumes 12 credits per second of video generation.

Credits are consumable units Runway provides with its plans. Each subscription tier offers a specific number of credits that users can access every month to generate videos. The lowest paid tier, Standard, offers 625 credits for $12 (roughly Rs. 1,034) a month. The most expensive tier, Unlimited, is priced at $76 (roughly Rs. 6,550) a month and offers unlimited credits to users.

Notably, the Gen-4 family of video generation models come with several upgrades over the previous generation. Runway says that with just one reference image, the models can generate consistent characters across different lighting conditions, locations, and camera angles.

The Gen-4 model can also use text prompts provided by a user to construct a scene with different camera angles, including close-ups and wide-angle side profiles. The models also improve upon the generation of real-world physics and motion. This means elements such as movements, object dropping, glass shattering, and wind effect will be more realistic.