OpenShorts is an open-source video clipper that turns long interviews, podcasts, and livestreams into captioned vertical clips.
If you record a 40-minute interview, paste its YouTube link or upload the file. OpenShorts transcribes the conversation and suggests stretches that can stand on their own. It can stack the host and guest in a vertical frame so both stay visible. Open a clip in the editor to move the cut, correct the captions, or adjust the framing before downloading it. See the YouTube clipping workflow.
Start with the MIT-licensed OpenShorts GitHub repository. The setup guide calls for Docker Compose and at least 8 GB of RAM. A compatible NVIDIA GPU is optional and can speed up processing. The self-hosted software adds no watermark or usage cap. You supply the machine and any API keys you use.
Transcription and video rendering run locally. If you use Gemini, it can receive transcript text and sampled frames, or the source footage when it contains no speech. You can use a local model through Ollama for selecting moments from a transcript. With only a local language model, the app uses its standard face-tracking crop and can't pick moments from silent footage. See hardware requirements and what stays local.
OpenShorts Cloud covers hosting and the Gemini key. The free plan processes up to 60 minutes of your first video, then gives you 20 source minutes per month. Free exports have a watermark and are available for seven days. Paid plans remove the watermark and let you connect accounts for scheduled posting.
| Cloud plan | Monthly price | Source video per month |
|---|---|---|
| Starter | $12 | 100 minutes |
| Creator | $29 | 300 minutes |
| Pro | $59 | 750 minutes |
Processing that 40-minute interview uses 40 minutes of the plan, regardless of how many clips it produces. The Starter plan could cover two interviews of that length in a month. See OpenShorts Cloud pricing and limits.
OpenShorts
Ready to try it? Open the official website.