Sora (OpenAI)
OpenAI's video app, shut down on April 26, 2026 (the API followed on September 24). What the group made with it and where members went next.
On this page
Sora was OpenAI's text-to-video app. It is gone: OpenAI announced on March 24, 2026 that it was closing Sora, switched off the website and phone app on April 26, 2026, and ended the Sora API on September 24, 2026. OpenAI said the Sora research team would move to world-simulation research for robotics. There is no Sora product to sign up for today.
If an old guide or a chatbot tells you to "make it in Sora," use one of the tools listed below instead.
The short history
- Fall 2025. OpenAI launched Sora 2 with a separate app. The group traded invite codes through October and November 2025, and members used it on the web at sora.chatgpt.com. A common question was whether it could make clips longer than about 8 seconds.
- December 2025 to February 2026. Members were still making clips: shared Sora videos to remix in December 2025, and in February 2026 a member asked whether short party-invitation clips needed a paid plan. That use was real; the shutdown came later.
- March 25, 2026. The builders group shared the news of the shutdown, and a member urged everyone to download their Sora videos.
- April 26, 2026. The app and website closed. OpenAI's help center pointed users to an export page and said data would be deleted after the last export window.
- September 24, 2026. The API, which some third-party video sites used behind the scenes, closed too.
If you still need videos you made in Sora, search OpenAI's help center for "Sora discontinuation" to see whether any export window is left.
What the group made with it
- Short fundraising clips.
- Holiday posts and event invitations.
- Educational clips for children. One member made educational videos for five or six months before moving to Grok in May 2026.
The lesson that carries over
The most useful thing the group learned on Sora applies to every video and image model: describe what the camera should see. Clothing, setting, lighting and what people are doing get far more accurate pictures of a Chabad House crowd than labels do. For example:
"Women in modern modest dresses, celebrating around a Shabbos table with candles and challah."
"A family in smart casual clothes lighting a menorah in a living-room window at night."
The fuller discussion, with examples, is in Prompting AI Image Tools to Depict Your Community.
Where members went
As of October 2026, in the order the group mentions them:
- Grok: the most common replacement for short clips; recommended "for vids" in the shluchos group in October 2026.
- Kling (kling.ai): suggested repeatedly for animated clips longer than 10 seconds. Kling 3.0 came out in February 2026; the site now leads with Kling 4.0.
- Google's video models: Veo 3.1 and the newer Gemini Omni, in the Gemini app and Google Flow with a Google AI subscription (Gemini Omni).
- Seedance (ByteDance): suggested in May 2026; version 2.5 is current and also appears inside CapCut.
- Personal agents such as Tab, which a member said makes one- to two-minute videos (October 2026). See Personal AI agents.
For the full roster, including editors and avatar tools, see Video Creation and Video Editing Tools.