OpenAI Launches Sora, an AI Video Generator

OpenAI has publicly released a new tool—Sora. Users can create videos from text prompts, “animate” images, extend existing clips, and fill in missing frames.

Our holiday gift to you: Sora is here. https://t.co/UhdmYuGHtT pic.twitter.com/ljoruQsfO0

— OpenAI (@OpenAI) December 9, 2024

The Sora neural network was introduced in February, but initially, only a few could use it. It is now available to users in the US and “most countries worldwide” with Pro and Plus subscriptions.

Exact launch dates for Europe, the UK, and some other countries have not been specified.

The released version is called Sora Turbo. It can generate video clips ranging from five to 20 seconds in various resolutions.

During a live broadcast on YouTube, OpenAI CEO Sam Altman demonstrated the Blend feature, which merges two scenes together.

“This is an early version of Sora. It is not perfect and makes mistakes. However, we believe it is at a stage where it is truly useful for enhancing human creativity,” noted OpenAI technical specialist Will Peebles.

Access to Sora is limited. Video creation requires credits, which are gradually spent depending on the video’s resolution and duration. Plus and Pro subscriptions provide 1,000 and 10,000 credits per month, respectively. Videos in 480p resolution cost 20-150 credits, while 1080p costs 100-2000.

Sora credits vary by resolution and duration, with ChatGPT Plus and Pro plans providing 1,000 and 10,000 credits respectively, resetting monthly without rollover

— Sora credits are required for video generation, with costs depending on resolution and duration: 480p videos cost… pic.twitter.com/4KZrgOfqCA

— Tibor Blaho (@btibor91) December 9, 2024

Credits do not accumulate—when new credits are issued, the old ones are annulled.

In November, several testers uploaded the not-yet-officially-released Sora to the Hugging Face platform in protest.

In August, it was revealed that Alibaba is working on Tora, a video generation tool based on OpenAI’s solution.