What should Sora users do now that the app is closing?

OpenAI announced on March 24 that it is discontinuing Sora, both the consumer app and the API, and said dates would follow. They now have: the Sora app and web experience go dark on April 26, 2026, and the Sora API, the paid service developers call from their own software, stops on September 24, 2026. After the deadlines pass, OpenAI says it will permanently delete user data. It has not yet decided whether it will offer one last export window afterwards; if it does, users will hear by email.

That leaves about three weeks for most people. The good news is that the work is simple if you do it in order.

Step 1: Export your library now

OpenAI's guidance is to download your content before the cutoff, and your library itself offers an export for both your videos and your images. Do it this week rather than on the last day, when servers may be busy and a failed download leaves no time to retry. OpenAI's help page on the Sora discontinuation has the current instructions; check it before you start, since details can change.

Save the files to two places, for example your computer and an external drive or cloud folder. A single copy is one accident away from being lost.

Step 2: Save your prompts and settings, not just the videos

A video file shows the result, but the prompt, the text description you typed, is what lets you make something similar elsewhere. Sora will not carry your prompts to another tool, so copy the ones that worked into a document before the app closes. A simple log makes each one reusable:

File name	Prompt (as typed)	Length and shape	Notes
harbor-dawn.mp4	Slow aerial shot of a fishing harbor at dawn, mist on the water, gulls circling	10 s, landscape	Second attempt; first was too dark
cat-astronaut.mp4	A ginger cat in a small spacesuit floats past a window showing Earth	5 s, portrait	Used a reference photo of our cat

Rows are illustrative. Note which videos include real people, such as yourself or friends, and where you have posted them. If a clip is shared on social media, keep the original file too; reposts are often compressed.

Step 3: If you build on the API, plan the switch

Developers have longer, but code changes take time. List every place your software calls Sora, what it sends (prompt, length, size) and what it does with the result. Then pick a replacement model, run the same few prompts through it, and compare quality, speed and price before rewriting anything. Keep the old and new paths side by side until the new one is tested.

Try it: Open your Sora library today and count the videos you want to keep. Export the three you care about most first, open each downloaded file to confirm it plays, then copy their prompts into a log like the one above.

Step 4: Decide what you actually need

Before choosing another tool, write down what you used Sora for. Short social clips, storyboards for a film idea, product mock-ups and family fun call for different features. Three questions narrow the field quickly: do you need sound, how long must each shot be, and do you need the same character to appear across several shots?

Which video tools can replace it?

Several companies offer text-to-video generation, and the field changes month to month, so test with your own prompts rather than relying on rankings. The most direct option from a large lab is Google's Veo. Its current version, Veo 3.1, was introduced in October 2025 and works inside Flow, Google's filmmaking tool. Google says Veo 3.1 improves sound, follows prompts more closely and handles turning images into video better than Veo 3. Some of its features are also coming to Google's Gemini API, which matters if you are a developer replacing Sora's API.

Flow's tools map well onto common Sora habits. "Ingredients to Video" lets you supply several reference images to keep a character, object or style consistent. "Frames to Video" creates the motion between a starting image and an ending image. "Extend" continues a clip from its final second, so you can build shots of a minute or longer. Here is how one of the prompts above could be adapted:

Prompt

Reference images: [photo of our ginger cat], [photo of a small white spacesuit]
Scene: The cat, wearing the spacesuit, floats slowly past a round window. Earth is visible outside, blue and bright. Soft hum of a spacecraft, a faint purr.
Camera: Steady, eye level, 8 seconds.

Notice what changed. The reference photos do the work that a long description used to do, and the prompt adds sound, since Veo 3.1 generates audio.

Whichever tool you try, compare it on the points in your Step 4 list, check its terms on who owns the output and how your uploads are used, and look at pricing per second of video. Then rerun two or three of your saved prompts. If the results are close enough, you have your replacement; if not, your prompt log means trying the next tool costs only a few minutes.