OpenAI Launches Sora 2 Video Generation Model
OpenAI released Sora 2, its next-generation video generation model with improved world simulation capabilities.
OpenAI released Sora 2, the successor to its video generation model first introduced in February 2024. The company describes the original Sora as a "GPT-1 moment for video" — the point where video generation began demonstrating emergent behaviors like object permanence through scaled pre-training compute. The Sora team has since focused on training models with more advanced world simulation capabilities, culminating in this second-generation release.
The launch marks a significant step in the competitive landscape for AI video generation, where OpenAI now fields an upgraded model against offerings from Google, Runway, and others. For developers and creative teams evaluating generative video tools, Sora 2's emphasis on world simulation — the ability to model how objects and environments behave realistically — signals a shift from pure visual fidelity toward more physically coherent outputs.