If your faceless video pipeline still has a step that calls OpenAI’s Videos API, that step has been pointing at a removed model since September 24, 2026. If the workflow has no error alert, it may have failed quietly for six weeks.
OpenAI shut Sora down in two stages: the app and website first, the developer API five months later. This post covers what changed and when, who it affects, and what to use instead, with every date and price checked against official pages on November 9, 2026.
The short answer
- Sora is gone in both forms. The web and app experience ended April 26, 2026. The Sora 2 models and the Videos API were removed from OpenAI’s API on September 24, 2026, with no replacement listed.
- We found no OpenAI page that lists video generation in any ChatGPT plan. Treat it as unavailable.
- For generated clips, Google Flow is the closest swap. Everyone gets 50 Flow credits a day; Google AI Plus adds 200 a month for $4.99.
- For faceless videos about software or workflows, a voice track over real screen recordings, or a Remotion template, never depended on Sora in the first place.
What changed, with dates
| Date | What happened | Source |
|---|---|---|
| March 24, 2026 | OpenAI notified developers that the Sora 2 video models and the Videos API were deprecated | OpenAI API deprecations page |
| April 26, 2026 | Sora’s web and app experiences ended | OpenAI support replies on OpenAI’s developer community forum, September 2026 |
| September 24, 2026 | sora-2, sora-2-pro, their dated snapshots, and the Videos API removed from the API. The replacement column is empty |
OpenAI API deprecations page |
All checked November 9, 2026.
What we could and couldn’t verify:
- Export and credits. On April 28, 2026, an OpenAI support reply on the community forum said existing projects could be exported at
sora.chatgpt.com/sunsetand that remaining credits could be used with Codex. OpenAI’s help article on the discontinuation blocked our requests, so we could not confirm the final export deadline or what happens to stored data. If you had projects there, check that page while signed in. - Video in ChatGPT. The ChatGPT pricing page lists no video generation for Free, Go, Plus, or Pro. In two September 2026 forum threads, OpenAI support answered requests for video inside ChatGPT by saying they would pass the request along, with no timeline. Some third-party articles say video moved into ChatGPT. We found nothing official that supports that.
Who this affects
Sora app users. Anything stored in Sora is only reachable through the export route above, if it still works for your account.
Anyone with an automated pipeline. A Make HTTP module or an n8n HTTP Request node that called the Videos API is calling models OpenAI has removed, so it can’t return a video. Search your scenarios and workflows for sora and for OpenAI’s video endpoint, and look at run history since September 24. While you’re there, add an error alert: a route that emails you on Make costs no credits, and n8n error-workflow runs don’t count against your execution quota.
Channels built on Sora’s look. Any replacement model renders differently. If your channel’s style was “whatever Sora made”, the next 20 videos will look like a different channel. That’s a reason to anchor the format in something you control: the script, the voice, the layout. The same goes for Sora-based templates and courses you bought: the prompts may still be useful, but the video step has to be rebuilt.
What to use instead
Three routes cover what faceless creators used Sora for. They solve different problems, and many channels will end up using two of them together.
| Route | Replaces | Cost (official pages, checked November 9, 2026) | Best for |
|---|---|---|---|
| Google Flow (Veo and Omni models) | Generated clips and B-roll | 50 free Flow credits a day for everyone. AI Plus $4.99/month adds 200 credits; AI Pro $19.99/month adds 1,000; AI Ultra adds 10,000 ($99.99) or 25,000 ($199.99) | Scenes you can’t film or screen-record |
| Remotion template | The whole video, built from your text and data | Free for individuals and small companies | A repeatable format made from your articles |
| AI voice plus screen recording | The narrator and the visuals | ElevenLabs Free has 10,000 credits a month with no commercial license listed; Starter is $6/month, or $5/month billed annually, with a commercial license | Tutorials, software walkthroughs, workflow tips |
Google Flow: the direct swap for generated clips
Flow is Google’s video studio, and it runs on credits. Per Google’s Flow help page (checked November 9, 2026):
- Veo 3.1 Lite costs 10 credits per 4, 6, or 8-second clip at 720p, or 5 on Ultra.
- Omni Flash 1.2 costs 3 credits per second at 720p, for clips up to 10 seconds on any account.
- Veo 3.1 Fast and Veo 3.1 Quality retire on December 5, 2026. Don’t build a new pipeline around either.
The math for a solo channel: AI Pro’s 1,000 monthly credits buy about 100 Veo 3.1 Lite clips, or about 33 ten-second Omni Flash clips at 720p. The free 50 daily credits cover about five Lite clips a day. Two caveats from the same page cut those numbers: some features create two generations per request and each one is charged, and unused credits don’t roll over. Non-subscribers may also be unable to generate video between roughly 2 and 5 p.m. UTC.
Before you commit, check three rules on Google’s Flow help page:
- You must be 18 or older with your age verified, and Flow isn’t available in every country where Google AI Pro is sold.
- Every Flow output carries an invisible SynthID watermark. A visible watermark is a toggle, except for users in India, South Korea, and Vietnam, where it’s applied automatically unless you’re on the Ultra 20x plan.
- Google’s Terms of Service (effective July 30, 2026) say Google won’t claim ownership of original content you generate with its services.
If you need an API rather than an app, Google also offers Veo to developers. We haven’t priced that route here, so check Google’s developer pricing before rebuilding an automated step on it.
Remotion: a template that never changes model
Remotion renders video from a React component, so one template turns each new script, table, and screenshot into a finished vertical video. It’s free for individuals and small companies. Nothing in it depends on a video model, so a shutdown like Sora’s can’t break it. Our faceless short-form video guide walks through the template, including Remotion’s official agent skills for Claude Code.
AI voice over real screens: the cheapest proof
For a channel about tools and workflows, the strongest visual is the real screen: the setting, the error, the result. Record it with your operating system’s built-in recorder (Shift-Command-5 on a Mac, Windows + Alt + R through the Xbox Game Bar), write a 110-word script, and voice it with ElevenLabs. Use the Starter plan once you publish, since the Free plan lists no commercial license. Viewers searching “how do I set this up” want to see the exact screen, and a generated clip can’t show it.
Our opinion: for faceless channels in the AI-tools and workflow space, generated footage was always the garnish. Sora’s exit is a cheap moment to move the core of each video onto material you own, and keep Flow for the shots you can’t capture.
What to do this week
- Search every Make and n8n workflow for
soraand OpenAI’s video endpoint, remove those steps, and add an error alert.- If you had Sora projects, sign in and try the export page OpenAI support pointed to.
- Make one test video on two routes (five Flow clips on the free daily credits, one screen-recorded video with an ElevenLabs voice) and keep the one that fits your format.
Quick answers
Is Sora still available inside ChatGPT? We found no official OpenAI page listing video generation in any ChatGPT plan as of November 9, 2026. OpenAI support has treated video in ChatGPT as a feature request, not a current feature.
Can I still call the Sora API? No. OpenAI’s deprecations page lists sora-2, sora-2-pro, and the Videos API as removed on September 24, 2026, with no replacement model named.
Is Google’s Veo free to try? Yes, within limits. Flow gives every account 50 credits a day, which covers about five Veo 3.1 Lite clips. Unused daily credits disappear at the end of the day, and non-subscribers may be blocked from video generation at peak hours.
The full production workflow, from script to render to upload, is in How to Automate Faceless Short-Form Video Without the Slop. More video and voice guides live in the AI Video & Audio Production pillar, and Turn Every New Post Into Social Drafts You Approve shows how to generate a video script for every article automatically.