OpenAI has unveiled Sora 2, a new video and audio generation model that the company says is more realistic, physically accurate, and easier to control than older systems. The model can create complex scenes like Olympic-level gymnastics, paddleboard stunts, and skating tricks while following real-world physical rules.
Along with the model, OpenAI also introduced a new iOS app called sora, powered by Sora 2. The app works like a social platform, allowing users to create videos, remix others’ content, explore a personalised feed, and even take part in cameos. By analysing a short video of a person, the app can insert their likeness and voice into any generated scene with high accuracy.
OpenAI CEO Sam Altman announced the launch on social media, calling it a major step in creative technology. He said the app feels like a “ChatGPT for creativity” moment, making it easy and fun to turn ideas into results. Altman added that Sora could spark a “Cambrian explosion” of creativity, raising the quality of art and entertainment worldwide.
According to OpenAI, Sora 2 is also a big improvement in controllability, capable of following detailed instructions across multiple shots and supporting cinematic, realistic, and anime-style outputs. With its features and social focus, Sora 2 and the Sora app are expected to compete directly with platforms like YouTube and TikTok.
