Can ChatGPT make videos?
No — and the guides saying otherwise are out of date. Video generation lived in Sora, not in ChatGPT, and OpenAI discontinued the Sora web and app experiences on April 26, 2026. The Sora API follows on September 24, 2026. ChatGPT still creates and edits images, reads video you give it in some forms, and talks — but it does not generate video for you.
OpenAI published a discontinuation notice. The search results have not read it.
The help article is titled "What to know about the Sora discontinuation" and answers its own first question without hedging: the Sora web and app experiences were discontinued on April 26, 2026, and the Sora API will be discontinued on September 24, 2026. That is the whole answer to this question. Video generation was Sora's job; Sora's consumer surfaces are gone and its developer surface has a date on it.
The rest of the notice is housekeeping that matters if you used it: content you created can be exported from sora.chatgpt.com/sunset, and after any final export window OpenAI will permanently delete data associated with your use of Sora. Purchased ChatGPT/Sora credits are not stranded — they can still be spent on Codex.
Scope, stated plainly: we did not run ChatGPT to test this, and we do not hold a ChatGPT account. What we did was read OpenAI's own current documentation — the discontinuation notice, the Images article updated the day before we checked, and the full ChatGPT release notes — and find no video-generation capability documented anywhere in ChatGPT as of 21 August 2026. If OpenAI ships a replacement, this page is wrong the day it launches, which is why it carries a date.
Worth knowing why the search results disagree so loudly: this cluster is ~19,000 searches a month at almost no keyword difficulty, so it attracted a great deal of content — nearly all of it written between 2024 and early 2026, when the answer genuinely was yes. Those pages have not been revisited. Ours will be.
What ChatGPT does do with moving pictures.
The capability people conflate with video is alive and well. ChatGPT Images 2.0 is available on all tiers, generating and editing images from a prompt, from an uploaded picture, or through a selection tool that targets one region of an image. It handles aspect ratios, transparent backgrounds and text inside the image.
Paid tiers add "images with thinking", where the model plans before generating. One retirement to note in passing: the official DALL·E GPT retires on August 30, 2026 — image generation continues through ChatGPT Images, and user-created GPTs with image generation enabled are unaffected.
Details on our image generation page.
"Can ChatGPT make videos" and "can ChatGPT watch videos" get typed by different people with different problems, and they have different answers. Uploading a clip, summarising a YouTube link, or extracting what happens on screen is an input question, and it is answered separately on our video analysis page.
One adjacent fact worth carrying: OpenAI's newer voice model, GPT-Live-1, explicitly does not support video or screen sharing at launch — those remain with Advanced Voice Mode for eligible subscribers. So even the live-camera capability that once made ChatGPT feel video-aware is currently split across two voice experiences.
We are not going to speculate about OpenAI's reasoning; the notice does not give one. What the record shows is a pattern rather than a one-off. During 2026 OpenAI retired Sora's consumer apps, announced the end of the Sora API, deprecated Atlas (its browser, which stopped working on August 9, 2026), retired group chats, retired the DALL·E GPT, and removed a long list of models from ChatGPT — while consolidating agentic work into ChatGPT Work and Codex.
The practical lesson for anyone building a workflow on a vendor's side feature: the surface you depend on can be withdrawn with a help-centre article and a date. That is not a criticism of OpenAI specifically — it is the shape of this whole market right now, and it is the reason every claim on this site carries the date we checked it.
Where generation still happens.
We have not tested these and publish no quality comparison between them — they are named because they are the surfaces that generate video from a prompt today, not because we are recommending one. Check each vendor's current terms before building anything on it.
Frequently asked.
Quick follow-ups people search after this question.