OpenAI intends to release the new Sora model specialized in creating video clips with the assistance of artificial intelligence during this year.
In an interview with The Wall Street Journal, Mira Morati, the Chief Technology Officer at OpenAI, indicated that the current plan is to launch the Sora model for all users later in 2024, which could take several months.
The company offers a limited number of users the chance to try the Sora model through an invitation system, aiming to secure it for users and enhance its performance before a wide release.
Morati explained that the Sora model has been trained on content from the Shutterstock platform, from which OpenAI obtained a license to use. However, she refused to disclose any other data sources, mentioning that they are from “publicly available or licensed data.”
Currently, the Sora model can only create silent video clips and does not allow editing during the creation process.
Regarding this matter, Morati stated that OpenAI is working on adding sound to Sora videos and developing editing tools.
There is no information available on the fees that OpenAI will charge the public for using their new model, but Morati confirmed that OpenAI intends to price Sora similarly to their previous model, DALL-E 3, although the cost of Sora is higher according to her.
Microsoft, the main partner of OpenAI, is also expected to integrate the Sora model into its products through its intelligent assistant, Copilot, just as it did with the GPT and DALL-E models in the past.
