On Monday the US start-up Runway unveiled Gen 4.5, its new artificial intelligence model capable of generating high-definition videos from simple text descriptions. The model now tops the independent Video Arena ranking, ahead of Veo 3 from Alphabet and OpenAI's Sora 2 Pro, thanks to outputs seen as more realistic, fluid and faithful to the laws of physics and human movement. The benchmark is based on blind comparisons between generated videos, evaluated by voters with no information about their origin.
Having just a team of around 100 people, Runway is managing to compete with tech giants valued at several trillion dollars. Its CEO, Cristóbal Valenzuela, sees this as proof that innovation and agility can still prevail in a market dominated by a handful of giants. The start-up, founded in 2018 and worth $3.55bn according to PitchBook, appears on CNBC's Disruptor 50 list this year. It provides its solutions to a broad customer base ranging from production studios to independent creators.
The Gen 4.5 model, nicknamed "David" internally, will be accessible to all Runway customers by the end of the week via its platform, its API and selected partners. This launch ushers in a new phase of innovation for the company, which is banking on optimization and research to remain competitive. Runway aims to establish itself as a credible alternative to the AI giants and advocates for a more open ecosystem, shielded from an excessive concentration of generative technologies.
