OpenAI said it has no plans to release an API for Sora, saying its AI models can produce fairly realistic videos with text descriptions and reference images.
In an AMA with members of OpenAI's development team, Romain Huet, OpenAI's head of developer experience, said that the Sora API is not under consideration at this time. “There are no plans for a Sora API yet,” he wrote.
The reason may be a capacity issue. OpenAI was forced to shut down its Sora-powered video creation and editing suite application shortly after launch due to higher than expected traffic. OpenAI CEO Sam Altman apologized for X.
Sora, One Prompt, Cinematic AF pic.twitter.com/X8gzdes20M
— Alex Patrascu (@maxescu) December 14, 2024
“We grossly underestimated the demand for Sola,” he wrote. “It will take some time to get access for everyone. We are trying to find a way to do it as soon as possible!”
OpenAI restarted Sora sign-ups a few days ago.
The choice not to prioritize APIs for Sora could put OpenAI at a disadvantage compared to one of its main rivals, Google. Google released a restricted access API for its video generation model Veo in early December. Google announced this week that Veo 2, the successor to the Veo that made headlines for its surprisingly high-quality output, will get an API sometime in 2025.
AWS has an API for the recently released Nova Reel video model. Additionally, many startups focused on generated video provide APIs for their models. One company, Runway, claims that its API is used “by the world's largest consumer technology companies to reliably generate millions of videos for their users.”