# A free, open-source video clipper for your desktop.

How ClipMill turns long videos into short clips for free, what its open-source licence covers, and what to check before installing.

Published: 2026-10-15. Updated: 2026-10-15.

## From a recording to a finished clip

ClipMill is a free desktop video clipping tool for macOS, Windows, and Linux. Import a local video, find useful moments, review the suggestions, edit captions and framing, and export MP4 clips for Shorts, Reels, TikTok, or another destination.

It works especially well as a workflow for English video podcasts, interviews, and other recordings built around speech. The goal is to help you find and finish a clip. You still decide whether a suggested moment makes sense on its own.

## What free and open source mean here

The app is free to download. Local analysis, editing, and export do not require a ClipMill subscription. Processing uses your computer, so the practical costs are your hardware, disk space, electricity, and time. Optional cloud processing can introduce charges from the provider you choose.

ClipMill's source is published under **AGPL-3.0-only**. You can inspect the code and read the licence through the source link in the site footer. Bundled libraries and model weights retain their own licences. Check the actual licence terms when modifying or redistributing the software.

There is no need to compile the application just to try it. The [download page](/download) offers installers for the supported operating systems.

## The clipping workflow

1. **Bring a recording.** Choose a local video, or import a YouTube video you have permission to use.
2. **Find candidate moments.** Local speech models create a timed transcript. A local Qwen editorial model can propose and review clips; the built-in picker is available without that model.
3. **Review the context.** Check that the clip starts clearly, includes the point, and ends after the thought is complete. Keep, reject, or select a moment yourself.
4. **Finish the edit.** Trim the timing, correct caption wording, choose the framing, and check the sound.
5. **Export.** Save MP4 video with burned-in captions and optional SRT or WebVTT subtitle files.

For a concrete example, follow the [podcast clipping guide](/guides/podcast-clips).

## Choose the right download

- **Mac:** Apple silicon, such as M1 or newer, with macOS 14 or later. Intel Macs are not currently supported.
- **Windows:** Windows 10 or 11 on an Intel or AMD 64-bit computer. Windows support is currently in beta.
- **Linux:** x86_64 with a glibc compatible with Ubuntu 22.04 or newer. Choose an AppImage or Debian package.

Setup downloads the processing components and models. Local editorial AI uses MLX on Apple silicon and llama.cpp on Windows and Linux. A suitable Vulkan GPU can accelerate Windows and Linux; CPU processing is also available. Allow several gigabytes for model files and additional working memory. The Models screen shows the choices and requirements for your computer.

See the [installation guide](/install) for the current unsigned-installer prompts and setup instructions.

## When a local clipper is a good fit

Choose this workflow when you want to work with recordings on your own disk, keep control of the edit, and use open-source software. Local processing can continue offline after setup, which is useful when uploading a large source recording would be inconvenient.

Your computer's speed matters. A CPU-only machine may take many minutes to analyse a recording. The current speech workflow focuses on English. ClipMill also leaves uploading and publishing the finished clip to you; direct channel publishing is still in development.

Read [how local processing works](/privacy), check the [release notes](/releases), or [download ClipMill](/download) to try it with a recording you know well.