Sora 2

Sora 2 is OpenAI's media generation model for creating richly detailed, dynamic video clips from natural language prompts or images. It generates video with synchronized audio, supporting 16 and 20-second clips in portrait and landscape, with dialogue, sound effects, and background soundscapes. The model is tuned for exploratory work, giving quick feedback on tone, structure, and visual style, and supports iterative editing including video extension, character generation, and in-painting. Sora 2 and the Videos API are deprecated and scheduled for discontinuation on September 24, 2026, with no announced replacement model. It represented OpenAI's approach to generative video before the company stepped back from that capability in its platform.