# Sora 2 Sora 2 is OpenAI's flagship video and audio generation model that creates richly detailed, dynamic video clips with synchronized audio from natural language prompts or images. For higher quality outputs, see [openai/sora-2-pro](https://replicate.com/openai/sora-2-pro). ## About the model Sora 2 represents a major advancement in AI video generation, offering significantly improved physics simulation, realism, and controllability compared to previous systems. The model can generate videos with synchronized dialogue, sound effects, and background audio, making it ideal for creating cinematic content across multiple styles including realistic, cinematic, and anime aesthetics. ### Key capabilities **Advanced physics simulation**: Sora 2 understands real-world physics including gravity, momentum, buoyancy, and object permanence. When a basketball misses a shot, it bounces realistically off the backboard rather than teleporting into the hoop. Objects move and intera