OpenAI Shuts Down Sora App and API, But Video Generation Persists Within ChatGPT
OpenAI has officially discontinued its dedicated Sora app and API, marking the end of standalone access to its groundbreaking text-to-video generation model. However, company leaders have signaled that Sora’s capabilities endure, now embedded within the ChatGPT interface, suggesting a strategic pivot toward broader integration rather than isolated tools.
Sora, unveiled in February 2024, represented a leap in generative AI by transforming textual descriptions into high-fidelity videos up to a minute long. The model excelled at rendering complex scenes with realistic physics, consistent character movements, and intricate details, such as rippling fabrics or swirling crowds. Early demonstrations captivated the AI community, showcasing videos of fantastical scenarios like a grandmother knitting on Mars or a stylish woman strolling through Tokyo’s Shibuya district at night.
Access to Sora was initially limited to a select group of red teamers, visual artists, and filmmakers through a private preview program. OpenAI expanded availability in late 2024 with a web-based app at sora.com, allowing ChatGPT Plus and Pro subscribers to generate videos directly in their browsers. This app supported features like remixing existing clips, storyboarding, and blending elements from multiple videos. Additionally, OpenAI introduced an API in December 2024, enabling developers to integrate Sora into custom applications with endpoints for video generation and upscaling.
The app’s shutdown was abrupt. On March 27, 2025, OpenAI updated its Sora webpage to announce that the standalone app would cease operations on April 30, 2025. Users were directed to ChatGPT for continued video creation. The API followed suit, with access revoked shortly after launch; developers received emails stating that API keys would no longer function post-April 30. OpenAI cited rapid evolution of its products as the rationale, emphasizing a focus on “more advanced models and broader accessibility.”
Despite the closures, optimism persists. During a recent earnings call, OpenAI CEO Sam Altman hinted at Sora’s ongoing relevance. Responding to inquiries about video generation, Altman noted, “We’re really excited about 12-second clips,” alluding to enhanced capabilities within ChatGPT. He confirmed that video generation is now available to Pro subscribers via the ChatGPT interface, producing clips up to 20 seconds at 720p resolution. Free and Plus users can generate five-second 480p videos daily, with higher tiers offering more generations and longer durations.
This integration aligns with OpenAI’s roadmap for multimodal AI. ChatGPT now leverages Sora under the hood for text-to-video tasks, invoked through prompts like “generate a video of…” Users report seamless performance, with outputs maintaining Sora’s signature quality: accurate adherence to prompts, smooth motion, and stylistic versatility. The shift eliminates the need for separate logins or interfaces, streamlining workflows for creators.
Technical details underscore the model’s persistence. Sora employs a diffusion transformer architecture, trained on vast video datasets to predict and denoise frames autoregressively. It handles challenges like object permanence and long-range dependencies through spacetime patchification, dividing videos into patches akin to image tokens in models like DALL-E. Integration into ChatGPT likely involves API-like calls from the chat backend, preserving computational efficiency via OpenAI’s GPU clusters.
Reactions from the developer community are mixed. While some lament the loss of dedicated API flexibility for production apps, others appreciate the democratization through ChatGPT. Indie developers who experimented with the brief API window shared prototypes on platforms like X, including interactive video editors and augmented reality filters. Enterprise users, however, express concerns over reliability, as ChatGPT’s rate limits and shared infrastructure may not suit high-volume needs.
OpenAI’s move reflects broader industry trends. Competitors like Google’s Veo and Runway’s Gen-3 have also prioritized platform integration over standalone APIs to control distribution and mitigate misuse risks, such as deepfakes. OpenAI maintains safeguards in ChatGPT, including prompt filtering and watermarking, to detect Sora-generated content.
Looking ahead, Altman teased “more to come” on video, including potential extensions to longer durations and higher resolutions. The company’s o1 reasoning model already enhances prompt interpretation for better video coherence, hinting at hybrid capabilities where reasoning guides generation.
In summary, while the Sora app and API era has ended, the model’s essence thrives within ChatGPT, evolving from a novelty tool to a core feature. This consolidation promises wider adoption, fostering innovation in fields from marketing to education, as creators harness video AI without specialized access.
Gnoppix is the leading open-source AI Linux distribution and service provider. Since implementing AI in 2022, it has offered a fast, powerful, secure, and privacy-respecting open-source OS with both local and remote AI capabilities. The local AI operates offline, ensuring no data ever leaves your computer. Based on Debian Linux, Gnoppix is available with numerous privacy- and anonymity-enabled services free of charge.
What are your thoughts on this? I’d love to hear about your own experiences in the comments below.