Last updated: 11/18/2026Last verified: 2026-11-15

OpenShorts

OpenShorts Overview: Open-Source AI Video Clipping

4
0

Open-source AI clip generator that turns podcasts, interviews and livestreams into 3 to 15 vertical 9:16 clips with face-tracked reframing, burned-in word-level captions and AI hook titles. Self-host free under the MIT licence with Docker and your own keys, or use the hosted Cloud from $12/month.

Freemium

What is OpenShorts?

OpenShorts is an open-source AI clip generator that turns podcasts, interviews, webinars and livestreams into vertical 9:16 shorts. Its own homepage describes the pipeline as faster-whisper transcription with word-level timestamps, PySceneDetect scene-boundary detection and Google Gemini 3.1 Flash-Lite scoring to pick 3 to 15 moments of 15 to 60 seconds each, which FFmpeg then cuts and MediaPipe face tracking reframes with a YOLOv8 fallback. It runs either as a self-hosted Docker app under the MIT licence or as the hosted OpenShorts Cloud. This listing read the official site, the pricing page, the GitHub README and the GitHub API on 2026-11-15, but did not install or run OpenShorts, so clip quality and the vendor's speed figures were not measured here.

Key Features

  • Moment detection combines faster-whisper word-level transcripts, PySceneDetect scene boundaries and Gemini 3.1 Flash-Lite scoring to return 3 to 15 candidate clips of 15 to 60 seconds each.
  • Vertical 9:16 reframing uses MediaPipe face tracking with a YOLOv8 fallback, damped so the crop holds still instead of chasing every head movement.
  • Word-level captions are burned in with FFmpeg, and Gemini-generated filter chains add hook overlays, colour treatments and transitions.
  • Optional ElevenLabs dubbing into 30+ languages, using your own key, in Cloud or self-hosted.
  • Direct posting to TikTok, Instagram Reels and YouTube Shorts from the dashboard.
  • A YouTube studio tool generates thumbnails, ten title suggestions and descriptions with chapter timestamps.
  • Agents can drive it through the native MCP server at mcp.openshorts.app/mcp, a REST API with per-user keys and completion webhooks, or the pip-installable CLI (pip install openshorts).
  • Self-hosting runs the same pipeline on your own Docker host, so the source video stays on your infrastructure.
  • Imports a local file or a YouTube, Twitch, Kick or TikTok link.
  • Every Cloud plan includes the same pipeline with the GPU and the provider API keys covered.

Best For

Podcast and interview creators repurposing long recordings into short-form social video.Livestreamers who want automated highlights for TikTok, Reels and Shorts.Content teams that need reframing, captions and hook titles in a single run.Developers who want an open-source, automatable clipping pipeline with an API, MCP server and CLI.Creators who prefer self-hosting so the footage never leaves their own hardware.

Pricing

OpenShorts is freemium. Self-hosted is $0 under the MIT licence: run it with Docker and bring your own Gemini, ElevenLabs and fal.ai keys, with no watermark and no minute cap. OpenShorts Cloud clips your first video free up to 60 minutes, then gives 20 minutes a month with a watermark and no credit card; Starter is $12/month for 100 source minutes, Creator $29/month for 300 and Pro $59/month for 750, all without a watermark. Yearly billing is $120, $290 or $590. Extra minutes bought from the app are 60 minutes for $9 or 200 minutes for $25 and do not expire at the end of the month. Billing counts started source minutes, so a 40-minute video costs 40 minutes whether it yields 5 clips or 12.

Pros & Cons

Pros

  • Open source under the MIT licence, with a Docker deployment that keeps the source footage on your own hardware.
  • The stated pipeline covers transcription, moment scoring, scene detection, reframing, captions and hooks in one run.
  • Face-tracked 9:16 reframing and burned-in word-level captions remove a manual crop and subtitle pass.
  • Automation beyond the web UI: a native MCP server, a REST API with webhooks, and a CLI.
  • Cloud billing is by source minute, extra-minute packs do not expire monthly, and the free tier needs no credit card.

Cons

  • Self-hosting shifts cost to hardware and keys: Docker with 8 GB+ of RAM plus your own Gemini, ElevenLabs and fal.ai credentials.
  • The homepage states an 8-minute video takes 5 to 8 minutes on a typical CPU and about 50 seconds with a GPU; those are the vendor's own figures and were not measured for this listing.
  • Gemini 3.1 Flash-Lite is required for moment scoring, so a self-hosted run needs a working Gemini key and pays that model's rates.
  • AI-picked moments, hook titles and filter chains still need a human pass before publishing.
  • The homepage counters (45,000+ accounts, 32,900+ videos) are the vendor's own numbers; this listing did not install the app or clip a video, so clip quality was not tested.

Alternatives

ClipShip

A Windows and Apple Silicon Mac desktop app that clips locally without Docker or a Gemini key.

Vugola

Adds scheduling and publishing to five social networks on top of clipping, as a paid subscription rather than open source.

OpenCut

A free open-source editor for cutting the clips by hand instead of letting AI pick the moments.

Descript

Transcript-based editing for podcasts and screen recordings rather than automatic clip harvesting.

FAQ

Details

Platform

WebAPImacOSWindowsLinux

Features

  • Moment detection: faster-whisper transcribes with word-level timestamps, PySceneDetect finds scene boundaries, and Gemini 3.1 Flash-Lite scores the transcript to pick 3 to 15 clips of 15 to 60 seconds
  • MediaPipe face tracking with a YOLOv8 fallback reframes to 9:16, FFmpeg burns in word-level captions, and Gemini-generated filter chains handle hooks, colour and transitions
  • Optional ElevenLabs dubbing into 30+ languages, direct posting to TikTok, Instagram Reels and YouTube Shorts, and a YouTube studio for thumbnails, titles and descriptions
  • Agents can drive it over an MCP endpoint (mcp.openshorts.app), a REST API with per-user keys and webhooks, or the pip-installable CLI, and Docker self-hosting keeps the footage on your own hardware

Languages

en

Known limitations

  • Cloud bills by source minutes: the free plan clips the first video whole up to 60 minutes and then 20 minutes a month with a watermark, and Starter, Creator and Pro cover 100, 300 and 750 minutes
  • Self-hosting moves the cost to hardware and keys: Docker with 8 GB+ of RAM, your own Gemini, ElevenLabs and fal.ai credentials, and an 8-minute video takes 5 to 8 minutes on a CPU
  • The pipeline depends on Google Gemini 3.1 Flash-Lite for moment scoring, so a self-hosted run needs a working Gemini key and pays whatever that model costs
  • The GPU speed figure (about 50 seconds for an 8-minute video) and the homepage account and video counters are the vendor’s own numbers, and this listing did not install the app or clip a video

Official pricing source

https://www.openshorts.app/pricing

Rate This Tool

Related Tools