What Is OpenShorts?
OpenShorts is a free, open-source app that turns long videos into vertical 9:16 clips. The core is MIT-licensed, so you can self-host it with Docker at no cost, with no watermark and no monthly cap. There is also a hosted version at openshorts.app with a free plan and paid plans from $12 a month.
Independent guide. AIClipMaker.org is not OpenShorts and is not affiliated with it. Facts from the GitHub repository and openshorts.app, checked on November 16, 2026.
OpenShorts at a glance
| What it does | Finds moments in a long video, crops them to 9:16 with face tracking, adds subtitles |
| Price | Free to self-host; hosted free plan, paid from $12/mo |
| License | MIT (core); cloud/ folder under a separate commercial license |
| Source code | github.com/mutonby/openshorts, about 8,400 stars |
| Runs on | Docker and Docker Compose, 8 GB+ RAM; NVIDIA GPU optional |
| AI it uses | Google Gemini to pick moments (or a local model via Ollama), Whisper for transcripts |
What OpenShorts does
The project bundles three tools in one dashboard:
- Clip Generator. You upload a podcast, webinar or stream. Gemini reads the transcript and picks 3 to 15 moments, then the app crops each one to 9:16. It tracks faces, can stack two speakers, or put a screen recording over the presenter, and burns in subtitles.
- AI Shorts. Marketing videos with AI actors, made from a product description or a URL. This part needs paid fal.ai and ElevenLabs keys; the README puts it at roughly $0.65 to $2 per video.
- YouTube Studio. Title ideas, thumbnails and descriptions with chapter timestamps.
It can also post to TikTok, Instagram and YouTube through Upload-Post, and it ships an MCP server, so Claude or ChatGPT can ask it to clip a video.
How to install OpenShorts
You need Docker and a Google Gemini API key (free from Google AI Studio). Then:
git clone https://github.com/mutonby/openshorts.git
cd openshorts
docker compose up --buildThe first build downloads PyTorch, Whisper and other models, so it takes a while. Full walkthrough with GPU setup and common errors: OpenShorts GitHub install guide.
How to use it
- Open
http://localhost:5175. - Go to Settings and paste your Gemini key. Add fal.ai, ElevenLabs or Upload-Post keys only if you want those features.
- In Clip Generator, upload a long video you own.
- Wait. On a typical CPU, an 8-minute video takes 5 to 8 minutes.
- Review the clips, then download them or publish them.
Self-hosted or hosted?
| Self-hosted | Hosted (openshorts.app) | |
|---|---|---|
| Price | Free (MIT); you pay for your machine and API use | Free plan; $12, $29, $59 a month |
| Watermark | None | On the free plan only |
| Setup | Docker, model downloads, your own keys | Sign in with Google |
| Speed (8-min video) | 5 to 8 min on CPU | About 50 s, per the README |
| Your files | Stay on your server | Processed on their servers |
When OpenShorts is not the right fit
If you don't want to run Docker or manage API keys, look at hosted tools instead. AIClipMaker.org picks the moments with AI, adds captions and reframes to 9:16 with face tracking, with nothing to install, and you can start free. For more options, see OpenShorts alternatives and OpenShorts vs OpusClip.
FAQ
Is OpenShorts free?
Yes, if you run it yourself: the core is MIT-licensed, with no watermark and no monthly cap. You pay for your own machine and any AI APIs you choose. The hosted version at openshorts.app has a free plan with a watermark. Details: is OpenShorts free?
Is OpenShorts open source?
Yes. The core app is under the MIT license. One folder, cloud/, holds the billing and hosted-service code and is under a separate commercial license; you never need it to self-host.
Who makes OpenShorts?
It is an independent open-source project on GitHub (mutonby/openshorts). It is not made by OpusClip. AIClipMaker.org is a separate site and is not affiliated with it.
Do I need a GPU?
No. The default Docker setup runs on CPU, at roughly 5 to 8 minutes per 8-minute video according to the README. An NVIDIA GPU makes it much faster.
Is there an OpenShorts app to download?
No desktop installer. You either run it with Docker from the GitHub repository or use the hosted website.