OpenAI is now launching Sora 2, according to a recent announcement post. Sora 2 is the next generation of the company’s AI video and audio generator, promising more realistic, physically accurate, and controllable results. Unlike previous models, which often “cheated” with physics, Sora 2 can generate more believable simulations.
Sora 2 can create videos of advanced movements—like gymnastics sequences, backflips on a paddleboard, and figure skating triple axels—while generating synchronized sound, dialogue, and effects. It can also insert people or objects from real-life videos, and recreate both their appearances and voices with high precision.
There’s also a new iOS app called Sora being launched. The app allows you to create and share AI videos, remix other people’s clips, and use the cameos feature to put yourself (or your friends) right into AI-generated scenes. If your cameo is used in a remixed clip, you’ll be counted as a co-creator (thus allowing you the power to remove said clip).
OpenAI emphasizes that this launch is focused on responsible use. The feed will encourage creation over passive consumption, and users will also be given tools to personalize their experience and minimize addictive scrolling.
At launch, the Sora app will only be available in the US and Canada. Sora 2 is available on the web. Both require an invitation from OpenAI.
This article originally appeared on our sister publication PC för Alla and was translated and localized from Swedish.