Sora, an AI-driven video generation model introduced earlier this year, now has a faster version called Sora Turbo. This updated model, capable of creating realistic videos from text, is available to ChatGPT Plus and Pro users.
Sora Turbo allows users to create videos up to 20 seconds long with resolutions up to 1080p. It supports various aspect ratios, including widescreen, vertical, and square. Users can either create new videos from text prompts or upload their own assets to blend, remix, or extend content. The platform also includes a storyboard tool for precise frame-by-frame inputs and community feeds showcasing recent creations.
Subscribers to the Pro plan can access higher resolutions, longer durations, and greater usage limits. Pricing tailored to different user needs is expected to roll out next year.
Metadata
Videos generated by the model come with metadata that identifies their origin, and visible watermarks are included by default. The platform has also implemented measures to prevent harmful content, including blocking deepfakes and child abuse material. At present, uploads of people are restricted, though this may expand as safeguards improve.
While Sora Turbo is faster and more advanced than its predecessor, it still faces challenges, such as generating complex actions and maintaining realistic physics. Developers aim to address these issues while working to make the technology more affordable.
Sora is included in Plus subscriptions at no extra cost, offering up to 50 videos per month at 480p resolution. This rollout provides an opportunity for users and society to explore the model’s potential while shaping norms for its responsible use.