Sora app

OpenAI's Sora app hit No. 1 on the US App Store three days after its September 2025 launch. About seven months later, OpenAI closed it.

What it was

The Sora app was OpenAI's social app for making, remixing and watching AI-generated videos, built on Sora 2, the company's video and audio generation model.

OpenAI's launch post described a customizable feed and a feature called cameos. After a short one-time video and audio recording to verify their identity, users could drop themselves or their friends into generated scenes.

TechCrunch called the feed TikTok-style and the app OpenAI's own TikTok competitor.

OpenAI had already released a Sora web product at sora.com for ChatGPT Plus and Pro users on December 9, 2024.

What happened

The Sora app launched on September 30, 2025, as an invite-based iOS app in the US and Canada. Making videos was free, within limits OpenAI's launch post said were "subject to compute constraints."

It became the No. 1 app on the US App Store on October 3, TechCrunch reported. Bill Peebles, OpenAI's head of Sora, posted on October 9 that it had reached a million downloads in under five days.

On October 16, OpenAI paused videos resembling Martin Luther King Jr., saying his estate had asked for it after some users generated "disrespectful depictions" of him, TechCrunch reported.

Copyrighted characters such as SpongeBob and Pikachu drew a backlash too, and OpenAI changed its policy for rights holders from opt-out to opt-in, according to TechCrunch.

On October 30, OpenAI started selling extra video generations on top of the free ones through Apple's App Store, Engadget reported.

Disney announced on December 11, 2025, that it had agreed to become the first major content licensing partner on Sora.

Downloads peaked in November at about 3.3 million for the month, then fell 32% in December and 45% in January, according to Appfigures estimates that TechCrunch reported. By late January the app sat at No. 101 among free apps on the US App Store, and TechCrunch wrote that the Disney news "hasn't increased Sora installs or consumer spending."

On March 24, 2026, OpenAI announced it was discontinuing the Sora app.

Disney found out less than an hour before the public, according to TechCrunch's summary of a WSJ investigation, and the deal died with the app.

The app and the Sora web experience were discontinued on April 26, 2026, according to OpenAI's help center. OpenAI shut down the API on September 24, 2026, and its model page says there is "no one-to-one replacement API."

Why it failed

The Sora app cost far more to run than it earned while its audience shrank, TechCrunch reported, citing Appfigures and the Wall Street Journal, and OpenAI said it was putting its compute toward its highest-value uses.

What OpenAI said

OpenAI's shutdown notices gave no reason. The Sora account's farewell post, as TechCrunch embedded it, began with "We're saying goodbye to the Sora app."

In a statement to CBS News that day, OpenAI pointed to compute tradeoffs. It said it was "prioritizing the highest-value uses" of its compute and that the Sora research team "continues to focus on world simulation research to advance robotics."

Sam Altman, OpenAI's CEO, gave his reason in an interview with the Mostly Human podcast, saying OpenAI needed to "concentrate our compute and our product capacity into these next generation of automated researchers and companies," as TheWrap reported on April 2, 2026.

Three days after launch, Altman wrote that people were "generating much more than we expected per user, and a lot of videos are being generated for very small audiences." On October 30, 2025, Peebles wrote that "the economics are currently completely unsustainable," Engadget reported. Both remarks came months before the shutdown, not as reasons for it.

What reporting showed

In a January 2026 report on Sora's struggles, TechCrunch's Sarah Perez wrote that videos of characters like SpongeBob and Pikachu "drove adoption" before the switch to opt-in, and that many users didn't want others, even friends, using their likeness.

"Without familiar faces and with limits on using commercial IP, people's interest in Sora seems to have faded," she wrote.

Her colleague Amanda Silberling wrote on the day of the announcement that "there was not sustained interest in an AI-only social feed."

The same piece put the app's lifetime in-app purchase revenue at about $2.1 million, by Appfigures' estimate. She doubted compute was the deciding factor, finding it "hard to imagine that the Sora app's computing demands tipped the scales that much" for a company already operating at a huge loss. The app was "perhaps too much of a liability to keep around if it wasn't even growing," she wrote.

Five days later, TechCrunch's Connie Loizos summarized a Wall Street Journal investigation that weighed compute more heavily. She wrote that, per the Journal, Sora's worldwide users peaked at around a million and fell to fewer than 500,000, while the app was "burning through roughly $1 million every day." Loizos summed up its finding as "Sora was a money pit that nobody was using," with Altman making the call to "kill Sora, free up compute, and refocus."

According to TheWrap, the Journal reported that the shutdown was part of OpenAI's effort "to redirect computing resources and top talent to productivity tools for enterprise and individual users."

What later analysis found

Constellation Research analyst Liz Miller judged Sora on quality for business use, not on OpenAI's decision to close it. She had already left it off a shortlist released in February 2026 and wrote on April 1, 2026 that "it wasn't even a close call."

Miller called Sora "a potential liability within an enterprise content creation process," with output that "lacked quality and was often unusable for final production."

The stage it reached

The Sora app never left the introduction stage. I'd file it as hype that never turned into growth, in the same stage where Humane's AI Pin died.

The product life cycle guide describes introduction as the launch period, when awareness is low, sales are small and costs are at their highest. Sora had plenty of awareness. Small sales and high costs fit.

Its pricing changed a month after launch, another sign of the stage.

Nothing points to growth. Appfigures estimates that TechCrunch reported show downloads and spending shrinking within months, even after the Disney deal.

It wasn't a decline either. The decline stage is when usage keeps falling for several quarters while a substitute grows, and Sora's fall lasted a few months before OpenAI withdrew it.

The lesson

The lesson I take from the Sora app is that when every use of a product burns compute, early demand is a cost before it is proof.

I'd judge a product like this on what an active user costs against what they pay, and on whether they come back once the novelty wears off. Downloads measure neither.

Write down the numbers that would make you stop before you launch, and put cost among them. Altman did publish a stop rule on launch day, but it measured satisfaction, not cost. If most users, looking back after six months, didn't feel Sora had made their lives better, he wrote, OpenAI would make significant changes, and "if we can't fix it, we would discontinue offering the service."