OpenAI CTO says Sora "could be a few months" away from public launch

OpenAI is gearing up to release its highly anticipated text-to-video AI model, “Sora,” to the public later this year, as confirmed by Mira Murati, CTO of OpenAI and former CEO (for two days, if you know, you know), in an interview with The Wall Street Journal. 

This announcement comes after the model made waves earlier this year with its ability to transform text prompts into jaw-dropping 60-second videos. This captured widespread attention within the tech community. Also, if you already didn’t know, a minute of OpenAI’s Sora video may take over an hour to generate. 

Interestingly, the debut of “Sora” coincided with Google’s unveiling of Gemini 1.5; some say Sora is an unfinished product but was revealed earlier due to Gemini 1.5. A few days later, to Sora’s preview, Microsoft admitted that OpenAI’s Sora would come to Copilot.

While initially limited to 60-second videos, “Sora” has showcased remarkable results, although access has been restricted to OpenAI thus far. According to Murati, this is set to change, with “Sora” slated for public release in the coming months. 

The interview shed light on the training data utilized for “Sora,” revealing a blend of publicly available videos and licensed content from platforms such as Shutterstock. However, specific inquiries regarding the incorporation of YouTube videos remained unanswered, hinting at the intricate complexities of AI model training.

Here is the interview.