⚡ Quick verdict
- Sora is gone — do not pay for it. OpenAI closed the Sora app and website on April 26, 2026, and removed the Sora API on September 24, 2026. There is no official way to generate a new Sora video today, free or paid.
- What it was: Sora 2, launched September 30, 2025 — text-to-video with synchronized audio, a TikTok-style social feed, and a Cameo feature that put your verified likeness into generated scenes.
- What it did best: short, cinematic, realistic-looking clips with dialogue, sound effects, and ambience generated in the same pass as the picture.
- What it cost while live: free in the app with daily limits; standard access with ChatGPT Plus ($20/month); the higher-quality Sora 2 Pro with ChatGPT Pro ($200/month).
- Use instead: Veo 3.1 for realistic clips with sound, Runway for creator control, Kling for cinematic styles — compared in detail below.
We may earn a commission if you buy through our links, at no extra cost to you.
If you landed here hoping to try Sora, here is the honest answer first: you can't — not anymore. OpenAI's video generator had one of the most talked-about launches in AI, then one of the quietest exits. The app that once topped download charts closed in April 2026, and the API followed five months later. Any website still selling "Sora access" or "Sora invite codes" in November 2026 is selling something that does not exist.
That makes this review different from most tool reviews on this site. It is a record of what Sora actually was, what it did well, what it cost, and — most usefully — what to use now that it is gone. If you made videos with Sora while it was live, we also cover how to get your old clips back. Everything here is dated November 2026 and drawn from OpenAI's own help documentation, its developer deprecation schedule, and independent reviews published while Sora was running.
The short answer: Sora is discontinued
Sora went from launch to shutdown in about seven months. That speed is the single most important fact in this review, because most Sora content on the internet still describes it as a live product. It is not. Here is the timeline, verified against OpenAI's help documentation and developer deprecation pages in November 2026:
Sora timeline: launch to shutdown
- February 2024
- OpenAI first shows Sora, a text-to-video model, with demo clips that reset expectations for AI video.
- December 2024
- The first Sora ships to ChatGPT Plus and Pro subscribers as a web tool.
- September 30, 2025
- Sora 2 launches with a redesigned model, an invite-only iOS app in the US and Canada, the sora.com website, a social feed, and Cameo.
- October 6, 2025
- The sora-2 and sora-2-pro models reach OpenAI's API for developers.
- December 2025
- Disney agrees a reported three-year licensing deal — reportedly worth around $1 billion — letting users generate more than 200 Disney characters in Sora.
- March 24, 2026
- OpenAI announces Sora will be discontinued and notifies API developers of the shutdown.
- April 26, 2026
- The Sora app and website close for everyone. ChatGPT Plus and Pro subscribers lose integrated Sora access the same day.
- September 24, 2026
- The Videos API and the sora-2 and sora-2-pro models are removed, with no replacement model named.
Two details in that timeline deserve emphasis. First, OpenAI named no successor — the deprecation notice lists the models as removed with no recommended replacement, which is unusual. Second, the shutdown was total: consumer app, website, and API all went, rather than the product being folded into ChatGPT. One October 2026 report said OpenAI was considering bringing the model back through a startup, but OpenAI has not confirmed anything. Treat any "new Sora" offer as unverified until OpenAI itself announces it.
What Sora was
Sora was OpenAI's text-to-video system: you described a scene in plain language, and it returned a short video clip that looked like real camera footage of that scene. The first version, shown in February 2024 and released that December, produced silent clips up to 20 seconds long and immediately became the benchmark every other video generator was measured against.
The product most people remember is Sora 2, the September 2025 relaunch. It was not just a better model; it was a different product idea. OpenAI rebuilt Sora as a social app as much as a generator: a TikTok-style feed of AI clips, a standalone iPhone app, remix tools, and a feature called Cameo that let you insert a verified version of yourself into any scene. For a few months it worked — the app spread fast on invite codes, and Sora 2 set the standard for what "good" AI video looked like.
Then, in March 2026, OpenAI announced the wind-down. The reasons were never fully spelled out in public: the app was reportedly expensive to run, and the social feed raised moderation and likeness questions that never fully went away. Whatever the cause, the result is what matters for anyone searching today: Sora is a finished product, not a current one. This review judges it as it was — and points you at the tools that inherited its crown.
What Sora 2 could do
Sora 2's headline improvement over the first version was sound. Earlier text-to-video tools made silent clips; you added music and voiceover afterwards. Sora 2 generated dialogue, sound effects, and ambient sound in the same pass as the picture, so a clip could arrive as a finished little scene — a character speaking a line, a door slamming, street noise underneath — rather than as silent footage waiting for an edit. That single change is why Sora 2 clips felt so much more "real" than its rivals at the time, and it is also why every serious competitor since has copied it.
The core generation modes were straightforward. Text-to-video was the default: type a prompt describing the scene, camera movement, lighting, and action, and get back a clip. Image-to-video let you start from a still image and animate forward from it, which was the practical way to keep a character or setting consistent across several shots. You could also extend or remix an existing clip — take a generation you liked and push the action further, or restyle it. Clips were short by design: most generations ran 10 to 20 seconds, with the higher-quality Pro model supporting longer cuts in some configurations. Exact limits changed more than once during the product's life, so treat any single number with caution.
Resolution depended on the tier. The standard model generated at up to 720p, while Sora 2 Pro — the experimental, higher-quality model — could reach 1080p. The app supported both landscape and vertical formats, which mattered because the feed was a vertical, phone-first experience. None of this was a full video editor: Sora made clips, not finished videos. If you wanted a two-minute video, you generated several clips and stitched them together elsewhere — a workflow the app never really tried to replace.
The app, the feed, and Cameo
The Sora app is what made Sora 2 famous — and what made it strange. OpenAI did not just ship a better model; it shipped a social network for AI video. Open the app and you got a vertical feed of clips other people had generated, which you could watch, like, and remix into your own versions. That feed is also why Sora 2 spread so fast: every clip was an advertisement for the product that made it.
Cameo was the feature that got the most attention at launch, and the most scrutiny. You recorded a short verification video of your face and voice, and Sora built a likeness model from it. Then you could drop yourself into any scene you prompted — as the main character, as a bystander, as a version of yourself in another time or place. You decided who else could use your likeness, and you could revoke access later. In December 2025, Disney's reported licensing deal extended the idea to fiction: more than 200 Disney characters became generatable inside Sora, which turned the feed into a stream of clips starring characters no other video tool could legally offer.
Cameo was also the feature that drew the loudest objections. Letting anyone generate video of a real person's face — even with consent controls — is a deepfake machine with a friendly interface, and regulators noticed. The app was blocked in the United Kingdom, Switzerland, and the entire European Economic Area throughout its lifetime, which cut off a large part of the audience a consumer social app normally needs. Face moderation on inputs was strict by default: uploading likenesses of real people outside the verified Cameo path was generally blocked. Those guardrails were real, but they never fully settled the argument, and the likeness question followed Sora until the day it closed.
Access to the app was invite-only at first. You needed a code from an existing user or a push-notification invite from OpenAI, and codes were scarce enough in the first weeks that they were traded and talked about like concert tickets. An Android version followed about two months after the iOS launch, and the website at sora.com offered the same generation tools in a browser. By early 2026 the invite system had loosened — and then, in April, the app closed for everyone at once.
How access worked — and what Sora cost
There was never a standalone Sora subscription, which confuses people to this day. You got Sora through one of three doors, and which door you used decided what you could generate:
Sora access while it was live
- The free app tier
- Generate in the Sora app for free, with daily limits tied to OpenAI's available compute. Core features included — text-to-video, audio, and Cameo — but capped generations.
- ChatGPT Plus — $20/month
- Standard Sora 2 access with more headroom than the free app tier, plus the rest of ChatGPT Plus.
- ChatGPT Pro — $200/month
- The higher-quality Sora 2 Pro model, higher resolutions up to 1080p, and the largest generation allowances. Pro subscribers also got priority when compute was tight.
- Developers (API)
- Usage-based pricing per second of video — reportedly around $0.10 per second at 720p for the standard model, and $0.30 to $0.70 per second for Pro depending on resolution. Roughly $1 for a 10-second standard clip.
Those numbers are historical, not current — nothing about them can be bought today — and they moved during the product's life. Early coverage described different credit allowances than later coverage, and at least one pricing roundup reports that a free tier was removed in January 2026, months before the shutdown. If you are comparing old articles and the figures disagree, that is why: Sora's pricing was adjusted several times in a short life, and the sources captured different snapshots. The figures above are the ones reported most consistently across independent pricing trackers in late 2026; check the official site for anything current, because OpenAI's ChatGPT plan prices are the only part of this story that still exists.
One warning that matters more now than it did while Sora was live: anyone selling Sora access, Sora invite codes, or a "Sora 2 web" account in late 2026 is not selling the real thing. OpenAI closed the app in April 2026 and removed the API in September 2026. There is no code, account, or backdoor that restores it. Third-party sites claiming to run Sora 2 in a browser cannot be running the real model — the only Sora output left in the world is video people exported while it was live. If a site asks you to pay for Sora, close the tab.
What Sora did better than its rivals
Judged as it was in late 2025, Sora 2 was the best AI video generator most people had ever used — and the gap was visible in the output, not just in the marketing. Four strengths explain why.
First, the picture itself. Sora 2's clips looked like footage. Motion was smoother and more physically believable than the competition at the time: people walked without their legs dissolving, cameras moved without the scene warping, objects stayed the same shape from one frame to the next. It was not perfect, but the failure rate on ordinary scenes was visibly lower than Runway or Pika at the same moment, and that is what made the feed so addictive.
Second, the sound. Synchronized audio in the same pass was Sora 2's real technical lead. Dialogue matched lip movement well enough to pass at a glance, sound effects landed on the action, and ambient sound gave scenes a sense of place. Rivals were still bolting audio on afterwards or shipping silent clips. The difference was not subtle: a Sora clip with a character speaking a line felt like a scene, while a silent clip from a competitor felt like a demo. Every major rival has since added native audio — Google's Veo 3 made it the headline feature — which tells you how important Sora's lead was.
Third, Cameo made video personal in a way no rival attempted. Other tools could animate a photo of you; Sora let you star in anything, with your own face and voice, under consent controls you set yourself. It turned video generation from a production tool into a form of play — put yourself in a movie trailer, in a historical scene, in a joke with a friend — and that playfulness is a large part of why the app spread. No surviving tool has copied Cameo directly, partly because the moderation and likeness questions it raised are genuinely hard.
Fourth, it was easy. You did not need to learn camera terminology or negative prompts or seed values. You described a scene the way you would describe it to a friend, and the model handled the rest. The app's remix button made iteration one tap: see a clip you like, remix it, make it yours. For casual creators and first-time users, that low floor mattered more than any advanced control.
Where Sora fell short
Sora 2 was the best of its moment, not a finished product — and the gaps mattered more the harder you pushed it. The first limit was length. Clips topped out around 10 to 20 seconds, with longer cuts only on the Pro model in some configurations. That is fine for a social post and useless for a scene: nobody cut a short film out of 15-second pieces without a lot of stitching, and the joins showed. Every text-to-video tool of the era had the same ceiling, but Sora's cinematic quality made the ceiling more frustrating, because the clips looked good enough to want more of.
The second limit was physics and fine detail. Sora 2 was much better than Sora 1 at keeping the world consistent, but hands-on reviews at the time still caught the familiar failures: hands with the wrong number of fingers in close-ups, text on signs and screens that dissolved into convincing-looking gibberish, and complex multi-person interactions that fell apart halfway through a clip. Crowd scenes and fast action were the weakest cases. The practical rule was the same as with every video model: the simpler the scene, the better the result.
The third limit was control. Sora was a prompt-and-pray tool: you described what you wanted and took what came back. There was no timeline, no keyframes, no precise camera control, and no way to say "keep everything about this clip, but change only the jacket." Remix and extension helped, but they were blunt instruments next to a real editor. Professionals who needed a specific shot found Sora charming and unusable in the same session. Runway, which built its whole product around control and editing, was the better tool for that job even while Sora was live.
Then there were the limits that had nothing to do with the model. The app never launched in the UK, Switzerland, or the EEA, so a large audience simply could not use it. The free tier's daily limits shrank when compute was tight, so heavy days meant waiting. And the social feed, for all its virality, raised a question OpenAI never fully answered: what happens when millions of people can generate realistic video of anyone who consents — and how confident can you be that the consent is real? Those are the questions that tend to end products, and in Sora's case, they may have helped.
Sora vs the alternatives you can use today
Sora is gone, but the job it did is not — and three tools now split the audience Sora once held. If you want the closest thing to what Sora 2 felt like, start with Veo. If you want control, start with Runway. If you want cinematic style on a budget, look at Kling. Our detailed head-to-heads cover two of these pairings: Sora vs Veo and Runway vs Sora. The short version is below, and our best AI video generators roundup ranks the full field with current pricing and free tiers.
The best Sora replacements in November 2026
- Veo 3.1 (Google)
- The closest replacement. Realistic footage with synchronized audio and dialogue in the same pass — the feature Sora pioneered, now Google's headline. Best for: ads, explainers, and social clips that need sound. Access runs through Google's plans and products, so check the official site for current pricing.
- Runway
- The creator's choice. Stronger control, editing tools, and an image-to-video workflow that professionals actually use — the areas where Sora was weakest. Best for: filmmakers, agencies, and anyone who needs a specific shot rather than a lucky one.
- Kling
- The cinematic option. Known for stylized, film-like motion and strong image-to-video, often at a lower price than the US tools. Best for: stylized clips, concept work, and creators who generate in volume. Check the official site for current plans.
The honest way to choose is by the clip, not the brand. Need a realistic talking-head ad with dialogue? Veo 3.1, because the audio arrives with the picture. Need to art-direct a product shot and iterate on it precisely? Runway, because control is the product. Need twenty stylized clips for a music video on a budget? Kling, because volume is where per-clip costs bite. And if the thing you loved about Sora was the social feed, none of the three copies that experience directly — the closest you get today is a tool with a public gallery and remix features.
One more comparison matters: price. Sora's per-second API pricing — about $0.10 per second at 720p — set the expectation that a 10-second clip costs about a dollar. Veo, Runway, and Kling all use credit systems that work out differently once you count rejects, and their headline prices are not directly comparable. Before you commit to any of them, generate the same prompt on two tools and count how many tries each one takes to give you a keeper. The cheapest tool is the one that gets it right on the second try, not the one with the lowest sticker price.
Can you still use Sora today?
No — not to generate anything new. The app and website closed on April 26, 2026, and the API was removed on September 24, 2026. There is no free tier, no paid tier, no waitlist, and no invite code, because OpenAI retired the entire product line. If you search for Sora today, most of what you find is either old content written while it was live, or third-party sites trading on the name. Both are worth treating with care.
If you made videos while Sora was live, you may still be able to get them back. OpenAI opened an export window for existing users after the shutdown was announced, so your old generations may be downloadable even though the generator is gone. Check OpenAI's help pages for the current status of that export window — export windows do not stay open forever, and OpenAI has said Sora data will be permanently deleted after the window closes. If your clips matter to you, download them now rather than assuming they will still be there next month.
Could Sora come back? One October 2026 report said OpenAI was considering bringing the model back through a startup, but that is a single unconfirmed report, not an announcement — and OpenAI has named no replacement model and confirmed no return. Until OpenAI says something official, the right assumption is that Sora is finished, and the right move is to pick a live tool.
Who should still care about Sora — and who should move on
You should still care if you made videos with Sora. Export them now, while the window is open, and keep local copies — they are the only Sora output that will ever exist, and some of it is genuinely good. You should also care if you study AI products: Sora is the clearest case yet of a beloved AI tool being shut down at the height of its popularity, and the lesson — that a model is not a business, and a feed is not a moat — will shape the next generation of video tools.
Everyone else should move on. If you are a creator choosing a video tool today, Sora's quality is no longer the benchmark — Veo 3.1, Runway, and Kling have all shipped since, and each of them beats Sora 2 on at least one dimension that matters. If you are a marketer planning a campaign, build on a tool with a future, not a memory. And if you are a developer, do not build on the sora-2 API: it no longer exists, and code written against it will fail.
There is one group in the middle: people who loved the Sora app's social experience and keep looking for its replacement. Our honest advice is to stop looking for an exact replacement and split the job instead — generate clips in Veo or Runway, and share them where your audience already is. The generation is what survived, in better tools, and that is where your time should go.
Verdict: was Sora worth it?
While it lived, yes — Sora 2 was worth the hype, with limits. For the price of a ChatGPT subscription you already had, or free in the app with daily caps, it produced the most realistic short video clips anyone had made with a text prompt, with sound included, and it let you put yourself in the scene. No rival at the time matched that combination. For casual creators, social posts, and prototyping ideas, it was the most fun video tool of 2025 — and the most impressive demonstration of where the category was heading.
But the honest verdict has to include the ending. Sora was a product you could not rely on: it launched invite-only, never reached Europe, changed its pricing, and then closed entirely after seven months, taking its API with it. Anyone who built a workflow around Sora had to rebuild it. That is the risk of building on a single company's experimental product, and it is a risk the alternatives carry too.
So: is OpenAI's video AI worth it in 2026? As a product you can use, there is nothing to buy — Sora is discontinued, and the answer is to pick Veo 3.1, Runway, or Kling instead. As a moment in AI history, it was worth every minute: Sora 2 showed the world what text-to-video could feel like, forced every rival to add synchronized audio, and proved that ordinary people want to star in their own generated videos. The tools that inherited its crown are better than Sora in most ways. They exist because Sora showed them what to build.
Frequently asked questions
Is Sora still available in 2026?
No. OpenAI discontinued Sora in 2026: the Sora app and website closed on April 26, 2026, and the Sora API — including the sora-2 and sora-2-pro models — was removed on September 24, 2026. There is currently no official way to generate a new video with Sora, free or paid, and OpenAI has not named a replacement model.
What happened to Sora?
Sora 2 launched on September 30, 2025, as an invite-only iOS app with a social feed and a feature called Cameo. OpenAI announced the shutdown on March 24, 2026, closed the consumer app and website on April 26, 2026, and removed the API on September 24, 2026. OpenAI opened an export window so existing users could download videos they had already made.
How much did Sora cost?
There was never a standalone Sora subscription. While it was live, the Sora app was free to use with daily generation limits, standard Sora 2 access came with a ChatGPT Plus subscription at $20 per month, and the higher-quality Sora 2 Pro model came with ChatGPT Pro at $200 per month. Developers could also call the API, reportedly for around $0.10 per second of video at 720p, with Pro costing more. Prices changed during the product's life, so treat these as historical figures — check the official site for anything current.
What is the best Sora alternative in 2026?
For most people, Google's Veo 3.1 is the closest replacement: it generates realistic video with synchronized audio and dialogue in the same pass. Runway is the strongest pick for creators who want fine control and an editing workflow, and Kling is a strong option for cinematic, stylized clips. Our Sora vs Veo comparison and our best AI video generators roundup compare all three in detail.
Can I get my old Sora videos back?
OpenAI opened an export window for existing Sora users after the shutdown was announced, so you may be able to download videos you generated while the app was live. Check OpenAI's help pages for the current status of the export window, and do not pay any third-party site that claims to sell Sora access or invite codes — those offers are not legitimate.
Is Sora coming back?
Not officially. One October 2026 report said OpenAI was considering bringing the model back through a startup, but OpenAI has not confirmed any return, and no official Sora product exists today. Until OpenAI announces something, treat any 'new Sora' offer as unverified and check the official site before believing it.