Sora is gone. OpenAI closed the Sora app and website on 26 April 2026, and its deprecations page lists the Sora 2 models and Videos API as shut down on 24 September 2026, with no recommended replacement. For most former app users, the closest replacement is Gemini Omni in the Gemini app. It makes video from a prompt, edits it in conversation and can put your own likeness in a clip. Veo 3.1 in Google Flow suits cinematic shots with native audio. Kling 4.0 and Seedance 2.5 both generate up to 30 seconds in one pass. If you're on a music team making release visuals, Visuals puts several of these models in one workspace built around the song.
We make Visuals, so weigh what we say about it accordingly. We've said where other tools are the better choice, and claims about them link to the vendors' own pages.
What happened to Sora
- 24 March 2026: OpenAI notified API developers that the Videos API and the Sora 2 models would be deprecated, per its deprecations page.
- 26 April 2026: the Sora app and web experience closed, per OpenAI's help article.
- 24 September 2026: the shutdown date for the Videos API,
sora-2,sora-2-proand their snapshots. OpenAI lists no recommended replacement.
If you still have videos in Sora, OpenAI's help article says to export them at sora.chatgpt.com/sunset as soon as possible; you'll get an email when the export is ready. If OpenAI offers a final export window, it will notify users by email first. After that, it will permanently delete Sora data. OpenAI also says purchased ChatGPT or Sora credits can still be used for Codex.
Sora alternatives at a glance
Checked on 2026-10-26 on the model makers' and platforms' own pages. Access and pricing vary by plan and region.
| What you used Sora for | Try | Why | Watch out for |
|---|---|---|---|
| Prompt to video, remixing, putting yourself in a clip | Gemini Omni (Gemini app, Flow) | Conversational edits, video-to-video, your own avatar | Needs a Google AI plan; video upload for edits isn't available in the EEA, UK, Switzerland and some US states (Google) |
| Cinematic shots with sound | Veo 3.1 (Google Flow) | Native audio, realism, prompt adherence | Flow is a filmmaking tool, not a feed |
| Long takes from many references | Kling 4.0 (Kling app, Runway, OpenArt) | 3–30 seconds in one pass, up to 15 references, keyframes | Not on Kling's lowest paid plan |
| Long takes with sound, heavy referencing | Seedance 2.5 (Jimeng, Doubao, and platforms such as Runway and Higgsfield) | 30 seconds in one pass, up to 30 images, 10 videos and 10 audio references | Direct access depends on region; most Western users reach it through a platform |
| Fidelity and editing control | Runway (Gen-4.5, Aleph, plus third-party models) | Edits existing footage; performance capture | Third-party models usually cost more credits than Gen-4.5 |
| Atmosphere, HDR, reshaping footage | Luma Ray3.2 | 1080p across modes, HDR, Modify Video up to 20 seconds | |
| Music videos, lyric videos and Canvas from your own track | Visuals | Starts from the song; Seedance, Kling, Veo, Gemini Omni, Runway and Luma in one workspace; lip sync, editing and consistent characters | Free plan makes treatments with stills, not a finished video |
| Many models, one bill | Aggregators: Higgsfield, OpenArt, Runway, Krea | Switch models without new accounts | Credits usually cost more than going direct |
| A social feed of AI video | Nothing equivalent |
The alternatives
Gemini Omni: the closest thing to the Sora app
Google's Gemini app now generates video with Gemini Omni:
- Edit in conversation: each turn can remove or replace objects, change the camera angle or modify the scene.
- Edit uploads: start from an uploaded video, or from up to five images.
- Your own likeness: create an avatar and type
@plus your Google username to put yourself in a clip. That's the nearest replacement for Sora's cameos, though there's no shared feed. - Sound: you can ask for audio with the video.
You need a Google AI plan on a personal account, and the feature isn't available to under-18s. Developers get the same model as Gemini Omni 1.1 Flash through the Gemini API. Google's announcement lists scene extension up to 40 seconds, first and last frames, 360p drafts and 4K upscaling.
Veo 3.1 in Google Flow
Flow, Google's creative studio, offers both Gemini Omni and Veo 3.1. Google describes Veo 3.1 as having native audio and strong physics, realism and prompt adherence. If you used Sora for single cinematic shots, start here.
Kling 4.0
Kling 4.0 is now Kling's headline model. Runway's Kling 4.0 page lists:
- clips of 3 to 30 seconds in one pass, with sound;
- up to 15 references (10 images and 5 videos);
- an opening frame, up to eight frames in between and a closing frame;
- edits to footage up to 30 seconds long.
On Kling's own plans page, paid plans start at $10 a month list for 660 credits, but Kling 4.0 starts on the next tier ($37 list). Kling 3.0 remains available for cheaper, shorter clips.
Seedance 2.5
ByteDance launched Seedance 2.5 on 31 July 2026. It generates up to 30 seconds of audio and video in one pass, with multiple rounds of extension. It takes up to 30 images, 10 video clips and 10 audio clips as references, and can edit to the timestamp. ByteDance released it on Jimeng AI and Doubao Pro, with API access via BytePlus ModelArk to follow. Outside China, it's sold through platforms including Runway, Higgsfield, Krea and Morphic.
Runway
Runway's own Gen-4.5 is available on all paid plans, and Runway also sells Kling, Veo and Seedance models. Its editing tools are the reason to choose it over a single-model app. Aleph changes existing clips, and Act-Two transfers an actor's performance onto a character. For how it compares with an aggregator, see Higgsfield vs Runway.
Luma Ray3.2
Luma's Ray3.2 generates at 1080p across every mode. It adds HDR generation (at 2× the credits of SDR), 360p drafts, and Modify Video, which reshapes existing footage up to 20 seconds long.
Aggregators
Higgsfield, OpenArt, Krea and Runway let you switch between several of the models above on one subscription. That's convenient while you're still deciding, but you usually pay more per clip than going direct. See Higgsfield alternatives for a fuller list.
If you used the Sora API
OpenAI names no replacement, so pick by model:
- Gemini Omni 1.1 Flash: the Gemini API in Google AI Studio, or Google's Gemini Enterprise Agent Platform.
- Runway: Runway Dev, its developer platform, which also carries third-party models such as MiniMax Hailuo 3.0. Seedance 2.5 and several Kling models are available through Runway MCP, per its changelog.
- Kling: Kling's own API service, plus resellers.
- Seedance 2.5: BytePlus ModelArk, plus resellers.
Budget time to re-tune prompts. Each model reads camera language, duration and audio instructions differently, and a prompt that worked in Sora rarely gives the same shot elsewhere.
Sora alternatives for music videos
A music video needs more than one good clip. It needs the same artist across dozens of shots, a performance that lip-syncs to the track and versions for YouTube, Shorts and Spotify Canvas. Single-model apps leave all of that to your editor.
For music teams, labels, agencies and artists, Visuals is the pick:
- The models from this page, in one workspace. On paid plans it hosts Gemini Omni Flash, Veo 3.1, Kling 3, Seedance 2.5, Runway Gen-4.5 and Luma Ray 3.2, and you choose a model per shot.
- Editing and lip sync. It edits footage you already have with Gemini Omni 1.1 Flash and Seedance 2.5, and lip-syncs to the track with Seedance and Kling.
- Characters instead of cameos. Where Sora had cameos, the Visuals Agent creates original characters for your song from your music or your ideas, and keeps them consistent across shots. You can also start from your own photos or images you have permission to use.
- The finish. You build the sequence and the release formats (lyric videos, Canvas, vertical cuts) around the track. AI handles the tedious parts; the creative calls stay yours.
The gauntlet is a 30-second Seedance 2.5 film made from a character sheet and one prop reference. It shows the reference-led approach.
Where Visuals isn't the answer: it has no social feed, and if you only need one model for non-music work, the model maker's own app is usually cheaper. The free plan makes three video treatments with stills, not a finished video.
Sora is a trademark of OpenAI. Gemini, Veo, Kling, Seedance, Runway, Luma, Higgsfield, OpenArt and Krea are trademarks of their respective owners.
How this was made: researched and written by the Visuals.fm team with AI assistance, then fact-checked by people against the sources linked in the text.