OpenShorts turns long videos into captioned, face-tracked 9:16 shorts you self-host with Docker, exposing an MCP server + API so agents can drive it.
Yes, OpenShorts is open source under the MIT license.
OpenShorts has a free tier plus paid plans. The open-source version can be self-hosted with no license cost; you pay only for your own infrastructure.
Yes, OpenShorts can be self-hosted (the source is available under the MIT license).
Open-source, self-hostable OpusClip alternative (MIT core, Python): turns long videos into vertical 9:16 shorts with AI moment detection, per-scene reframing modes (track / general / split / screencast), face tracking, word-level captions via faster-whisper, and AI voice dubbing. Ships a built-in MCP server (/mcp) and a REST API with completion webhooks so agents can drive it. Self-host free with Docker & Compose, or use the managed cloud from $12/month (the cloud billing directory is source-available under a commercial license, not MIT).
Short-form clips are a distribution channel, and OpusClip-style tools bill per minute for exactly the workflow a content-led GTM motion runs constantly: cut the webinar/podcast/demo into clips. OpenShorts makes that pipeline something you own — Docker on your own box, no per-minute credits. The GTM-specific hook is the MCP server plus REST API with webhooks: the clip step becomes a callable tool in an agentic content pipeline (transcribe, clip, caption, post) rather than a person in a web app. Honest caveat: it is open-core, so the managed cloud is not MIT, and self-hosted output quality tracks your GPU and the underlying models — verify a dub before it goes to an audience.