OpenAI announced the launch of Sora, a new artificial intelligence model capable of generating videos starting from textual descriptions. The announcement is part of OpenAI’s “ship-mas” release series, a 12-day event during which the company introduces new features and products. Sora is available via Sora.com to paid ChatGPT subscribers. The template offers several features, including generating videos from text, animating images, and remixing existing videos. ChatGPT Plus subscribers can generate up to 50 priority videos (1,000 credits) with resolutions up to 720p and duration of 5 seconds. Subscribers to ChatGPT Pro, the 229 euro per month plan launched last week, have unlimited creations available and up to 500 priority videos with resolution up to 1080p and duration of 20 seconds. The Pro plan also allows you to download videos without watermarks and run up to five generations at once.
Sora was teased by OpenAI in February, and the launch was confirmed today by YouTuber Marques Brownlee (MKBHD), who shared a preview based on his experience testing the model. During a live stream, OpenAI presented Sora’s new “Explore” page, which offers a AI-generated video feeds created by other community members. The company also highlighted the “Storyboard” feature, which allows you to generate videos based on a sequence of prompts, and the ability to turn photos into videos. OpenAI also showed off a “remix” tool that lets you edit Sora’s output with a text prompt and a function to “merge” two scenes via AI.
Videos generated with Sora will have visible watermarks and C2PA metadata to indicate that they were created with AI. Before uploading an image or video to Sora, users must agree to an agreement that prohibits uploading content that includes minors, violence, explicit or copyrighted material. Improper use of the uploaded media may result suspension or termination of your account. Aditya Ramesh, vice president of research at OpenAI, said during the live stream: “We know we are a major target. We want to prevent illegal use of Sora, but also balance that with freedom of creative expression. We know it will be a challenge continues and that we may not achieve perfection from day one. We are starting conservatively, so if our moderation isn’t perfect, please provide us with feedback.”
Users without a ChatGPT subscription will still be able to explore the feed of AI-generated videos created by other users. The model will be available in the US and several other countries starting today, but OpenAI CEO Sam Altman said the rollout in most of Europe and the UK may take longer. Sora’s release comes just a week after protest by a group of artists who, claiming to have participated in OpenAI’s alpha testing program, released information about the product, accusing the company of having exploited them for “unpaid research and development and public relations”.
