OpenAI announces updated Sora 2 video model

The Sora 2 video model has been announced by OpenAI, following a year after their introduction of Sora. This major update enables the platform to generate more realistic and physically accurate videos with upgrades to user control and editing.

Sora 2 offers better audio output with synchronized dialogue and addresses issues from earlier models like distortion and unrealistic motion. Plus, it can now also follow multi-shot instructions and keep visual consistency across different scenes.

Some of its supported styles include anime, cinematic, and realistic. Sora 2 will begin its rollout today, but is currently limited to the US and Canada.

For those curious, Sora 2 will be free with ‘generous limits’ for the time being so users can explore its features. However, ChatGPT Pro subscribers will get access to a higher-quality Sora 2 Pro model.

OpenAI also noted that they intend to release Sora 2’s API. Aside from this, we covered that they released a standalone, invite-only Sora app on iOS.

It has several restrictions, such as measures to prevent generating videos of public figures. Users may also allow or revoke permissions for others to make cameos using their likeness.