Sora 2

Sora 2, launched by OpenAI in September 2025, could generate physically accurate videos from text prompts, supporting multi-shot storytelling with synchronized dialogue, sound effects, and background music. Users could upload a portrait to create a consistent digital double. It excelled at simulating gravity, light, and collisions, delivering coherent and realistic visuals. However, its use was heavily restricted: initially limited to the US and Canada with an invite-only system, creating scarcity and scalping. Video generation required heavy computation, making it expensive and the API asynchronous, requiring polling. On April 26, 2026, OpenAI discontinued Sora 2's consumer app, developer API, web version, and the ChatGPT video feature, ending its short-lived availability. During its brief existence, Sora 2 appealed to filmmakers, marketers, and developers needing professional AI video, but it is now completely unavailable. It was never suited for those on a tight budget, in regions without stable access, or needing simple animations.