Last updated: 11/22/2026Last verified: 2026-11-21

PeakCut

PeakCut Overview: Self-Hosted AI Clipping for Videos and Live Streams

11
0

PeakCut is an MIT-licensed, self-hosted AI clipper for long videos and live streams: paste a YouTube, Twitch or Kick link or upload a file, and it transcribes the audio, asks an AI which moments would hold a stranger, cuts them to 9:16 with word-by-word captions and a hook, and can post to connected accounts. It runs on your own computer or server with your own AI key, and works with no key at all by picking clips on loudness.

Free

What is PeakCut?

PeakCut is a free, MIT-licensed, self-hosted clipping tool for long videos and live streams. Its README describes it as: Turn long videos and live streams into captioned vertical clips, ready to post. Free and open source: run it on your own computer or server, with your own AI key. You paste a YouTube, Twitch or Kick link or upload a file; it transcribes the audio, asks a chat model which moments would hold a stranger's attention, and renders each clip to 9:16 with word-by-word captions, a hook and one of three caption styles. It also records a stream into a rolling buffer and clips highlights while the stream is still running, keeps separate workspaces per creator, and queues or auto-posts to connected YouTube, Instagram and TikTok accounts. The GitHub repository is the product: the project publishes no separate site and no tagged releases.

Key Features

  • Link or file input: a YouTube, Twitch or Kick link, or an uploaded video file.
  • Transcript-based moment selection: transcribes the audio and asks a chat model which moments would hold a stranger's attention.
  • Vertical rendering: 9:16 clips with word-by-word captions, a hook, and one of three caption styles.
  • Live clipping: records a stream into a rolling buffer and clips as it happens, using loudness, chat activity and the transcript.
  • Workspaces per creator: channels, clip settings, posting rules and an activity log kept separately for each creator you clip.
  • Posting: connect YouTube, Instagram and TikTok, then queue or auto-post with spacing and daily caps.
  • Your own AI: Gemini, OpenAI, Grok, Groq, OpenRouter, Claude, a local model through Ollama, or any OpenAI-compatible endpoint; with no key at all, clips are picked by loudness.
  • Local dashboard: opens at 127.0.0.1:8765 with no sign-in, and on a shared server you can switch accounts on with PEAKCUT_ACCOUNTS=1.
  • Server deployment: DEPLOY.md covers three containers on one Linux server, the app, PostgreSQL and Caddy for HTTPS.

Best For

Creators repurposing long videos into captioned vertical clips for social platforms.Streamers who want to capture highlights while a broadcast is still running.Technical users who prefer running the clipping workflow on their own computer or server.People managing several creators who need separate channel settings, posting rules and activity logs.Users who want to choose their own AI provider or run a local model through Ollama.Creators trying automated clipping without an AI key, using loudness-based selection.

Pricing

The software is free and MIT-licensed: the LICENSE file opens with MIT License and Copyright (c) 2026 PeakCut contributors, and the README says: You pay the AI service directly for what you use. The repository lists no plans, seats or subscription. What you do pay for is whichever model endpoint you connect, plus your own hardware: DEPLOY.md suggests a Linux server with 8+ CPU cores, 16 GB RAM and 200+ GB of disk if several people will use it, and notes that transcription runs at roughly 10x real time per video on 4 cores. Those are the project's own planning figures, not a quoted price.

Pros & Cons

Pros

  • MIT licence: use, modify and redistribute under the licence terms.
  • Handles both uploaded or linked videos and live streams in the same app.
  • Clip selection, vertical rendering, captions and posting in one workflow.
  • Provider-agnostic: eight model options, including a local Ollama model and any OpenAI-compatible endpoint.
  • Works with no AI key at all, picking clips by loudness.
  • Runs on your own machine with a local dashboard, or on your own server with the bundled Docker Compose stack.

Cons

  • You operate it: Python 3.10 or newer and ffmpeg are required, install is a git clone, and there are no tagged releases to pin.
  • The first start downloads a speech model of about 480 MB, and Arabic needs a further model of about 107 MB.
  • Captions default to English: the README says it hears English by default, and that Arabic speech is understood but captions on the clips are in English.
  • On Windows, an install path whose full length exceeds 260 characters stops the speech model from loading.
  • One server only: jobs and live recordings run inside the app process, so DEPLOY.md says not to run a second copy against the same data.
  • Instagram posting needs a Meta app, and until Meta reviews it only accounts added as testers can connect; Instagram also fetches each clip from a one-hour signed link on your site.
  • Name collision: a separate commercial service at peakcut.ai, run by Nythral Studio, uses the same name. The repository's homepage field is empty and neither references the other, so this listing covers the MIT repository only.
  • Verified from the repository README, DEPLOY.md, LICENSE and GitHub metadata on 2026-11-21; not deployed or tested for this overview.

FAQ

Details

Platform

WebWindowsmacOSLinux

Features

  • Transcribes the audio, asks a chat model which moments would hold a stranger's attention, and renders each clip with captions in one of three styles
  • Live clipping records a stream into a rolling buffer and clips as it happens, using loudness, chat activity and the transcript
  • Workspaces per creator keep channels, clip settings, posting rules and an activity log separate, and clips can be queued or auto-posted to YouTube, Instagram and TikTok
  • Bring your own model: Gemini, OpenAI, Grok, Groq, OpenRouter, Claude, a local model through Ollama, or any OpenAI-compatible endpoint — or no key at all, with clips then picked by loudness
  • Runs as a local dashboard at 127.0.0.1:8765 with no sign-in, or on a server through the bundled Docker Compose setup with PostgreSQL and Caddy

Languages

en

Known limitations

  • It is free and MIT-licensed but you pay the AI provider directly for what you use, and it needs Python 3.10 or newer plus ffmpeg installed on your machine
  • The first start downloads a speech model of about 480 MB, and captions are English by default — Arabic needs a separate model download and the caption text stays English
  • One server only: jobs and live recordings run inside the app process, so a second copy must not be started, and the Docker deployment guide plans for 8+ CPU cores, 16 GB of RAM and 200+ GB of disk
  • Posting needs a one-time per-platform setup, Instagram connections only work for accounts added as testers until Meta reviews the app, and a Windows install path longer than 260 characters stops the speech model from loading
  • There are no tagged releases yet (install is git clone), and a separate commercial service at peakcut.ai by Nythral Studio uses the same name; the repository links to no website and the two do not reference each other

Alternatives

OpenShorts — Self-hostable under MIT with a hosted cloud tier, but its pipeline depends on a Gemini key
SupoClip — A hosted plan sold by source minutes on top of an AGPL-3.0 self-host route
ClipShip — A closed-source desktop app with a paid tier instead of a Python server you maintain

Rate This Tool

Related Tools