New Delhi: OpenAI has announced Sora 2, its newest video and audio generation model that aims to create content with better physics, realism and control than any system before it. The launch also comes with a dedicated iOS app called “Sora,” where users can generate, remix, and even insert themselves into AI-created videos through a feature called cameos.

The company describes Sora 2 as a moment similar to when GPT-3.5 changed language AI. Unlike earlier models that often bent reality to fit a prompt, Sora 2 tries to follow the rules of the physical world. A missed basketball shot, for example, now rebounds off the backboard instead of magically teleporting into the hoop.

What makes Sora 2 different

OpenAI says Sora 2 is built to handle complex actions that were previously impossible.

See Full Page