Sora by OpenAI

Sora by OpenAI is a generative video model that creates high-quality videos from text prompts, images, or short clips. Its primary purpose is to simulate realistic or stylized scenes, characters, and camera movements while maintaining temporal consistency and coherent visual storytelling. Sora is designed to help users prototype visual ideas, explore concepts, and produce detailed video content without traditional filming or animation workflows.

The model can generate videos up to around a minute in length with multiple characters, complex backgrounds, and dynamic actions, while preserving object permanence and scene continuity. It supports a range of visual styles, from photorealistic footage to animation-inspired looks, and can interpret nuanced textual instructions about mood, composition, and motion. Sora is also capable of extending existing videos, transforming still images into moving scenes, and simulating physically plausible interactions between objects and environments. Built on large-scale video and multimodal training, it is optimized for coherence across frames and fine-grained control over content.