“Sora” is an AI model from OpenAI which can turn text prompts into jaw-dropping 60-second videos and, now, OpenAI says it will become publicly available later on in 2024.
OpenAI first showed off “Sora” earlier this year, ironically on the same day that Google announced Gemini 1.5. The text-to-video AI model caught attention for its ability to create stunning videos based on text prompts, with some examples almost indistinguishable from actual video captures.
As it stands, OpenAI has said that “Sora” is limited to 60-second videos, but the results have been pretty incredible. So far, though, the only output we’ve seen has been from OpenAI, as the model hasn’t been publicly available. That’s soon to change, though.
In an interview with The Wall Street Journal, OpenAI’s Mira Murati confirmed that “Sora” is coming “definitely this year” and potentially in “a few months.”
The interview also confirmed that OpenAI aims to allow users to “edit” output by directly asking to fix issues in the videos. The model also won’t work with audio for the time being.
Where “Sora” got its training data was also a core part of the interview, with Murati confirming that “publicly available” video and licensed video from Shuttershock was used to train the model, but many specific questions, such as if YouTube videos were used, were left unanswered.
