Runway, a New York City-based startup, has announced the release of a faster version of its AI model. The Gen-3 Alpha Turbo is said to be 7 times faster than the original Gen-3 Alpha model. This new model will be rolled out with significantly lower pricing in the coming days. Runway’s CEO, Cristóbal Valenzuela, stated that users will be able to generate new videos in “real-time” with the Gen-3 Alpha Turbo, producing a 10-second video in just 11 seconds.
The decision to release a faster version of their AI model with lower pricing may be a strategic move by Runway. By offering faster video generation, the company hopes to increase overall usage and spending on their subscription plans or a-la-carte generation “credits” model. Currently, Runway offers different monthly subscription plans, each with a set number of credits required for each generation of still images or video. The Gen-3 Alpha model costs 10 credits for every 1 second of video generated, while the older Gen-2 model is priced at 5 credits per second. The oldest model, Gen-1, is the most expensive at 14 credits per second. It is speculated that the Gen-3 Alpha Turbo may be offered at around 7 credits per second or even as low as 5.
However, Runway has faced some controversy recently. A former Runway employee allegedly leaked a spreadsheet showing the company’s plans to scrape and train its AI models using copyrighted content from popular YouTube channels. The company has not yet commented on these allegations. Additionally, Runway is facing lawsuits from creators who claim that the company has violated their copyrights on still images. Critics like Ed Newton-Rex, founder of the nonprofit organization Fairly Trained, have called out Runway to disclose its training data set. It remains to be seen if these lawsuits will lead to disclosure of training data and potential legal ramifications.
Overall, Runway’s release of the Gen-3 Alpha Turbo showcases their commitment to staying ahead in the field of generative AI video models. By offering faster and more affordable video generation, they aim to maintain their position as a leader in providing highly realistic, Hollywood-quality AI video models. The future will reveal how the market and legal landscape will shape the development and use of such models.