Sora 2
Sora 2 was OpenAI's video and audio model, released Sep 30, 2025 with an invite-only social Sora app for iOS (Android in November 2025) and cameos of real people. The app closed on Apr 26, 2026 and the Sora 2 API on Sep 24, 2026.
Sora 2 was OpenAI's second video model, released on Sep 30, 2025 with "synchronized dialogue and sound effects" and a new social app for iOS called simply Sora. It let people make short videos from a prompt, remix others' videos, and put themselves or friends into scenes through "cameos" after a one-time recording, a feature later renamed "characters". It drew huge early interest, then criticism over deepfakes and copyrighted characters, and it ended with the rest of Sora: the app and website on Apr 26, 2026 and the API on Sep 24, 2026. OpenAI offers no video model as of Nov 6, 2026.
Quick facts
| Released | Sep 30, 2025 |
| Status | Discontinued: app and web Apr 26, 2026, API Sep 24, 2026 |
| App | iOS from launch, invite-only, US and Canada first; Android in November 2025 |
| Price at launch | "Initially be available for free, with generous limits" |
| Higher quality model | Sora 2 Pro, for ChatGPT Pro users on sora.com |
| API price (before shutdown) | sora-2 $0.10 a second; sora-2-pro $0.30 to $0.70 a second |
| Provenance | Visible watermark and C2PA metadata on every video |
What Sora 2 could do
OpenAI called Sora 2 possibly "the GPT-3.5 moment for video", against the original Sora's "GPT-1 moment" (OpenAI). Its main change was sound: dialogue and effects generated together with the picture. In the API, sora-2 and sora-2-pro made 16- and 20-second clips, 1080p only on Pro, and a video could be extended up to six times by 20 seconds each, to a maximum of 120 seconds (OpenAI).
The Sora app and cameos
The app had a customizable feed, remixing and cameos. To make a cameo, a person recorded a short video and audio clip once in the app, which verified their identity and captured their likeness (OpenAI). In OpenAI's Sep 30, 2025 launch video, the team showed the flow: a spoken audio challenge, a liveness check that asks you to turn your head, then a setting for who may use your cameo (only you, people you approve, mutuals or everyone). At first OpenAI blocked uploads of images with photorealistic people and all video uploads (OpenAI). The company Cameo sued over the name, and OpenAI renamed the feature "characters" after the court's ruling (TechCrunch). By Mar 23, 2026 OpenAI also allowed image-to-video with real people once the creator attested to their consent (OpenAI).
OpenAI described its feed rules on Feb 3, 2026: optimize for creativity, let people steer the ranking, and let parents turn off personalization and endless scrolling through ChatGPT's parental controls (OpenAI).
Regions and the Android app
The app launched invite-only in the US and Canada, and was also usable on sora.com after an invite (OpenAI). In the launch video OpenAI said each person let in from the wait list got four invite codes to bring friends. The Android app came out in November 2025, reached first place in the Play Store on launch day and saw more than a million videos made in its first 24 hours; four OpenAI engineers built it in about a month with Codex (OpenAI). The full list of later countries was not found on any OpenAI page read for this article.
Limits and money
OpenAI said Sora 2 would start free "with generous limits" and that its only plan was to let people pay for extra videos when demand outran compute (OpenAI). No official per-plan daily or monthly generation limit, and no price for extra videos, survives on OpenAI's pages; the Sora help articles were removed. Sam Altman wrote that people were making much more video per user than expected and that OpenAI would have to find a way to make money from it, including sharing revenue with rightsholders (Sam Altman). TechCrunch reported Appfigures' estimate of about $2.1 million in lifetime in-app purchases (TechCrunch).
Watermarks and content rules
Every Sora 2 video carried a visible watermark and C2PA metadata at launch, and OpenAI ran internal image and audio search to trace videos back to Sora. The rules blocked public figures except through their own cameos, sexual material, terrorist propaganda, self-harm promotion and music imitating living artists or existing works, and added limits for teens (OpenAI). By March 2026 many videos carried moving watermarks with the creator's name (OpenAI).
Controversies
- Deepfakes of the dead: TechCrunch reported that guardrails on public figures "were easy to evade", and that deepfakes of Martin Luther King Jr. and Robin Williams led their daughters to ask users to stop (TechCrunch). CBS News reported that OpenAI temporarily blocked videos using King's likeness (CBS News).
- Copyrighted characters: users made videos with Mario, Naruto and Pikachu, according to TechCrunch. Altman promised rightsholders more granular control, similar to the opt-in model for likeness, and acknowledged "the remarkable creative output of Japan" (Sam Altman).
The Disney deal
On Dec 11, 2025 Disney and OpenAI announced a three-year licensing agreement: Sora and ChatGPT Images could use more than 200 Disney, Marvel, Pixar and Star Wars characters (not actors' likenesses or voices), curated Sora videos would appear on Disney+, and Disney would make a $1 billion equity investment in OpenAI. It was to start in early 2026 and depended on final agreements. After the shutdown news, TechCrunch reported that no money appears to have changed hands before the deal collapsed (TechCrunch).
How Sora 2 ended
OpenAI announced on Mar 24, 2026 that it was saying goodbye to the Sora app; the app and web closed on Apr 26, 2026, and the Sora 2 models and Videos API on Sep 24, 2026 (OpenAI). Why it closed and what to use now are on Sora and Sora vs Veo.
Frequently asked questions
Is the Sora 2 app still available? No. The Sora app and sora.com were discontinued on Apr 26, 2026, and the Sora 2 API shut down on Sep 24, 2026.
What were Sora cameos? Cameos let a person record a short video and audio clip once in the Sora app to verify their identity, then appear in Sora 2 videos. OpenAI renamed the feature "characters" after Cameo's lawsuit over the name.
Was Sora 2 free? At launch OpenAI made the Sora app free with what it called generous limits, and gave ChatGPT Pro users the higher quality Sora 2 Pro model on sora.com. In the API, sora-2 cost $0.10 a second of video.
How long could Sora 2 videos be? In the API, Sora 2 made 16- and 20-second clips, and a video could be extended up to six times to a maximum of 120 seconds.
What happened to the Disney and OpenAI deal? Disney and OpenAI announced a three-year licensing deal and a $1 billion Disney investment on Dec 11, 2025. TechCrunch reported that no money appears to have changed hands before the deal collapsed with Sora's shutdown in 2026.
Sources
- Sora 2 is here, OpenAI, Sep 30, 2025
- Sora 2 System Card, OpenAI, Sep 30, 2025
- Launching Sora responsibly, OpenAI, Sep 30, 2025
- Sora update #1, Sam Altman, read Nov 6, 2026
- Disney and OpenAI agreement, OpenAI, Dec 11, 2025
- Shipping Sora for Android with Codex, OpenAI, Dec 12, 2025
- The Sora feed philosophy, OpenAI, Feb 3, 2026
- Creating with Sora safely, OpenAI, Mar 23, 2026
- OpenAI's Sora was the creepiest app on your phone. Now it's shutting down, TechCrunch, Mar 24, 2026
- OpenAI discontinues Sora, CBS News, Mar 24, 2026
- What to know about the Sora discontinuation, OpenAI Help Center, read Nov 6, 2026
- sora-2, OpenAI API docs, read Nov 6, 2026
- Video generation guide, OpenAI API docs, read Nov 6, 2026
- Deprecations, OpenAI, read Nov 6, 2026
- Introducing Sora 2, OpenAI, Sep 30, 2025 (video)
Moments
Introducing Sora 2
OpenAI
- Sora 2 was OpenAI's second video model, released on Sep 30, 2025 with "synchronized dialogue and sound effects" and a…
- OpenAI called Sora 2 possibly "the GPT-3.5 moment for video", against the original Sora's "GPT-1 moment" (OpenAI). Its…
- The app had a customizable feed, remixing and cameos. To make a cameo, a person recorded a short video and audio clip…
- The app launched invite-only in the US and Canada, and was also usable on sora.com after an invite (OpenAI). In the…
- Every Sora 2 video carried a visible watermark and C2PA metadata at launch, and OpenAI ran internal image and audio…
See also
- OpenAIOpenAI is the AI company behind ChatGPT, Codex and the GPT models. Since Oct 28, 2025 it has been OpenAI Group PBC, controlled by the nonprofit OpenAI Foundation, with Microsoft holding about 27%; CEO Sam Altman leads it, and it was last valued at $852 billion.
-
SoraSora is shut down. OpenAI's text-to-video model, previewed in February 2024 and sold from Dec 9, 2024, ended as an app and website on Apr 26, 2026 and in the API on Sep 24, 2026; OpenAI has no video product as of Nov 6, 2026. -
CodexCodex is OpenAI's coding agent: it reads a codebase, edits files, runs commands and tests, reviews pull requests and works in the cloud. As of Nov 6, 2026 it lives inside the ChatGPT desktop app, in the Codex CLI, in IDE extensions, in Codex Cloud and on mobile, and every ChatGPT plan includes some of it, Free and Go too.
-
Sora vs VeoOpenAI's Sora app and website shut down on Apr 26, 2026 and its API on Sep 24, 2026, so the choice for AI video is now between Google's Veo and Gemini Omni, Runway, Kling, Seedance, Grok Imagine and others; on Arena's text-to-video leaderboard of Nov 5, 2026 Google's Gemini Omni leads, and Sora 2 Pro had ranked just above Veo 3.1.
-
ChatGPT ImagesChatGPT Images makes and edits images from a prompt in any chat, on every plan; the current version, Images 2.6 since Nov 4, 2026, adds looping GIFs and saved reference images. Free gets limited images, Plus and Pro get images with thinking, and Pro is unlimited within abuse guardrails. -
GPT ImageGPT Image is OpenAI's family of image generation and editing models in the API, from gpt-image-1 (Apr 2025) to GPT-Image-2.6 Sunburst and Flare (Nov 4, 2026). The current models cost $8 per 1M image input tokens and $30 per 1M image output tokens, and power ChatGPT Images.