What it does
OpenShorts takes a long video, uploaded as a file or pasted as a YouTube link, and cuts it into vertical clips for TikTok, Instagram Reels and YouTube Shorts. It transcribes the video with faster-whisper, detects scene boundaries, and uses Google Gemini to score the transcript and choose between 3 and 15 moments of 15 to 60 seconds. Each clip is cut with FFmpeg and reframed to 9:16 with face tracking. Extras described on the site include word-level subtitles burned into the clip, AI-written hook text over the opening seconds, Gemini-generated video effects, dubbing into 30+ languages with your own ElevenLabs key, a thumbnail and title generator, and direct posting to the three platforms. A separate AI UGC generator writes a script and produces a lip-synced avatar video from $0.65 per video. It can be used in two ways. The MIT-licensed code runs in Docker on your own machine with your own API keys, so the source video stays on your infrastructure. OpenShorts Cloud is the hosted version. The site also lists a REST API, an MCP server for tools such as Claude, ChatGPT and n8n, and a pip-installable CLI.
Capabilities
- Picks 3 to 15 clip-worthy moments from a long video using Gemini
- Face-tracked 9:16 reframing, with a separate mode for group shots
- Word-level captions, AI hook text and optional video effects
- Voice dubbing into 30+ languages with your own ElevenLabs key
- Posts directly to TikTok, Instagram Reels and YouTube Shorts
- REST API, MCP server and CLI for automation
When to use it
- Repurposing podcasts, webinars, livestreams or interviews as short vertical clips
- Keeping source footage on your own machine by self-hosting with Docker
- Automating clipping from an AI agent or n8n workflow through the MCP server or API
Alternatives to explore
- Opus Clip
- Klap
- Vizard
- Submagic
