Sora 2 is a multimodal AI video generation service based on OpenAI's Sora 2 model that converts text prompts and optional reference images into realistic short videos (typically up to 20 seconds). It includes native audio synthesis (dialogue and environment), enhanced physics simulation for believable motion, and editing tools such as Remix and Reframe to refine outputs. Users pick resolution, aspect ratio and length, then generate, iterate, and download or share results. The platform uses a credits-based pricing model with tiered plans for creators, professionals and studios, offering private generation, no-watermark outputs on paid plans, and commercial licensing.