Sora Access Expansion encompasses the stages through which OpenAI made its Sora text-to-video model available to the public, from initial previews to large-scale releases and eventual shutdown. Developed by OpenAI, Sora used Artificial intelligence to generate short video clips from text prompts and could extend existing videos. The model was first shown publicly in February 2024, with a public release for ChatGPT Plus and Pro subscribers in December 2024, followed by a second generation in September 2025 that added social media features. The service was discontinued in 2026, marking the end of the Sora brand.
Background
Before Sora, several other text-to-video models existed, including Meta (AI research)'s Make-A-Video, Runway's Gen-2, and Google Veo. OpenAI had previously released DALL·E 3 in September 2023, extending its portfolio of generative models. Sora's development built on these advances, aiming to produce higher-quality and longer videos than prior systems.
Initial reveal and preview
The team named the model Sora after the Japanese word for 'sky' to signify its 'limitless creative potential'. On February 15, 2024, OpenAI first previewed Sora by releasing multiple high-definition video clips it had created, including an SUV driving down a mountain road, an animation of a 'short fluffy monster' next to a candle, two people walking through Tokyo in the snow, and fake historical footage of the California gold rush. The company stated it could generate videos up to one minute long. A technical report highlighted the training methods, and CEO Sam Altman responded to Twitter users' prompts with Sora-generated videos.
Public release and expansion
As of December 9, 2024, OpenAI gradually made Sora available to ChatGPT Pro and ChatGPT Plus users in the United States and Canada. Prior to that, access was limited to a small 'red team' of experts in misinformation and bias for adversarial testing, as well as a group of creative professionals for feedback. In February 2025, OpenAI announced plans to integrate Sora into ChatGPT, allowing users to generate videos directly from the chatbot.
Sora 2 and social media features
Sora 2 was unveiled on September 30, 2025, with an iOS app released simultaneously and an Android app following two months later. All videos generated included a visible, moving digital watermark to help viewers distinguish real from fictional content. The app was noted for its similarity to TikTok in style and features. On October 7, 2025, 404 Media reported that third-party programs capable of removing the watermark had become widespread.
Discontinuation
On March 24, 2026, OpenAI announced on X that it would discontinue Sora in both the mobile app and the API. The app was shut down on April 26, 2026, and the API on September 24, 2026. OpenAI's partnership with Disney, which included a licensing agreement allowing Disney characters within Sora, was also ending. The decision led The Register to label OpenAI a 'product-killer', following other tech companies like Google, Amazon Web Services, Broadcom, and Netscape.
OpenAI did not provide a specific reason in its shutdown notice, but reports linked it to computation shortages, cost pressures, and a shift toward core enterprise products. At its peak, Sora had about one million worldwide users, declining to fewer than 500,000, while operating costs were estimated at $1 million per day due to the computational demands of video generation.
Legal and copyright issues
In November 2024, a group of testers leaked an API key for Sora on Hugging Face, posting a manifesto protesting that Sora was used for 'art washing'. OpenAI revoked access three hours later and stated that 'hundreds of artists' had shaped development and participation was voluntary.
At launch, Sora 2 allowed copyrighted content by default unless copyright holders actively opted out. On October 3, 2025, OpenAI announced a future update would give copyright holders 'more granular control', but did not clarify whether existing content would be removed. On October 6, the chairman of the Motion Picture Association criticized OpenAI's copyright approach. On December 11, 2025, Disney announced a three-year, $1 billion investment in OpenAI, allowing Disney+ users to generate over 200 copyrighted characters on Sora 2, including from Disney Animation, Pixar, Marvel Studios, and Star Wars.
Capabilities and limitations
Sora's technology adapted that of DALL-E 3. According to OpenAI, Sora is a diffusion transformer - a denoising latent diffusion model with a transformer as its denoiser. Videos are generated in latent space by denoising 3D 'patches', then transformed to standard space by a video decompressor. Recaptioning augmented training data by using a video-to-text model to create detailed captions.
OpenAI trained Sora on publicly available videos and licensed copyrighted videos, without revealing exact numbers or sources. The company acknowledged limitations, including difficulty simulating complex physics, understanding causality, and differentiating left from right. To adhere to safety practices, Sora restricted prompts for sexual, violent, hateful, or celebrity imagery, and content featuring existing intellectual property.
Researcher Tim Brooks noted that the model learned to create 3D graphics from its dataset, while fellow researcher Bill Peebles said the model automatically discovered many physical rules from data.