Tell it what
to cut.
It cuts.
The AI-native, non-linear video editor. Native apps for Android, tablets, Chromebooks, and desktop — macOS, Windows, and Linux. A real multi-track timeline, keyframes, and true mp4 export — no account, no cloud required.
Multi-track editing, built for real work.
Stack as many video and audio tracks as you need. Scrub, zoom, split, trim, drag clips across tracks, group them, and drop keyframes with a bezier curve editor.
A full editor — and an AI that uses the same tools.
Trim silences, clip every shot with a face in frame, or apply a look across a whole clip. Type it into the in-app assistant and it drives the editor through the same tools — every bit of analysis running on-device, your video never leaving the device.
AI keep/remove
Prompt a clip — “cut the silences”, “keep only shots with a dog”, “delete every frame with my phone” — and the on-device analyzer (object, face & scene detection) marks ranges to keep or remove. Point it at a scrubbed frame to track that object. Batch across many clips at once.
Multi-track timeline
Stacked video + audio tracks that truly composite — top track over the ones below, with crossfades where a track's clips overlap. Scrub, zoom, split, drag across tracks, group, and long-press a clip edge to trim or re-extend its in/out point.
Keyframes & curves
Animate opacity, scale and volume with keyframes, and shape the ease-in/ease-out with a cubic-bezier curve editor.
Filters & audio
Brightness, contrast, saturation, sepia, hue, invert, grayscale and blur — plus per-clip volume, L/R pan and normalization, with presets.
Real mp4 export
On-device rendering with Media3 Transformer physically cuts removed ranges, bakes in effects, crop and aspect, mixes audio, and saves to your gallery.
Phone, tablet & Chromebook
One adaptive layout that reflows for portrait phones and large screens, with keyboard shortcuts and mouse + Ctrl-scroll zoom. Installs without a touchscreen.
Save & resume
Store the whole project as a portable .gilt file and pick up exactly where you left off — clips, edits, keyframes and all.
Undo, copy & paste
Deep undo/redo history, plus copy-and-paste for whole clips and for the effects on a clip — so a look you dialled in transfers in one move.
Batch the boring stuff
Select many clips, write one prompt, and run the AI across all of them at once. Group clips to select, move and edit them together.
One-tap background removal
Tap once: on-device subject segmentation drops a clip's background so the layer below shows through — live in the preview and baked into export. One more tap blurs every face to anonymize. No key, no cloud.
One-tap auto-captions
Tap Auto-captions and pick a style — clean subtitles or animated word-pop that grows each syllable as it's spoken. Transcription runs on-device (your audio never leaves the device); captions land in the lower third and burn into the export.
Crop, scale & rotate
Pinch, drag and twist with two fingers to size, place and rotate the selected clip — video, image or text — directly on the preview.
In-app AI assistant
A command bar: type what you want and an agent drives the editor through the same tools. Use any capable cloud model (as a text-only controller) or the fully local on-device LLM brain — no video is ever sent.
Generative object removal
“Remove my phone and make it look natural.” The object is detected and masked on-device, the masked frames are repainted (BYO Leonardo.ai key), and the result is a grouped run of clips the same length as the original.
Runs in the background
Analysis, generative removal and export keep running when you leave the app, with a live progress notification. Pause and resume analysis or generative jobs — or cancel any of them — right from the notification.
LUTs, shaders & pro filters
Drop a .cube LUT or a GLSL/ISF shader onto a clip — a shader's inputs become adjustable sliders right in the popup — and reach the whole FFmpeg/Frei0r filter ecosystem. Graded and rendered on-device.
Transitions & beat-sync
Overlap two clips to crossfade, or bake in wipes, slides and dissolves via FFmpeg xfade. Detect a track's beat grid on-device and cut or animate your footage in time with the music.
Generate media — free, no key
Text-to-image via Pollinations and text-to-video on Guillotine's free Hugging Face Space — no key, no account. Bring your own key for many more image, video and music providers. Only your prompt is sent, never your media.
Open to any AI (MCP)
A built-in MCP server (localhost:6274) lets any external AI, agent or plugin drive the editor through the very same tools — the base of an open plugin, LUT and shader ecosystem.
Free by default. Powerful when you plug in a key.
Tell the in-app assistant what you want and it drives the editor for you. Every analysis runs on-device — your video never leaves your device. Cloud AIs are controllers: plug in your own key and the provider only exchanges text (your instructions + the timeline), never your clips or frames. Prefer fully local? Use the on-device LLM brain. Keys are stored encrypted on-device.
Local
On-device silence detector. Works offline, zero configuration.
On-device vision
ML Kit face & object detection, fully offline — keep/remove by what's on screen.
Free generation
Pollinations makes images and Guillotine's free Space makes video — no key, no account. Only your prompt is sent, never your media.
Gemini
Drives the editor as a text controller via the MCP tools (no video sent).
OpenAI
Text controller via MCP; Whisper still handles cloud transcription if you choose it.
Anthropic
Claude drives the editor as a text controller via MCP (no frames sent).
OpenRouter
One key, many vision models via an OpenAI-compatible endpoint.
Groq
Fast text controller via an OpenAI-compatible endpoint.
xAI · Mistral
Grok and Pixtral vision, OpenAI-compatible. Every model id is editable.
Your footage never leaves your device.
No account to make, nothing to upload, no one strip-mining your media. Every frame is analyzed where it already lives — on your device — and stays there.
All on your device
Import, edit, analyze and export run locally — face, object and scene detection, silence and beat detection, captions and background removal. No account, no upload, no server of ours in the loop.
Cloud AI only sees text
Choose a cloud model and it acts purely as a controller: it receives your instructions and the timeline, and replies with edits. Your frames, your audio, your footage are never sent.
Yours, and open
Your content stays yours and so do the rights to it — no data harvesting on your media, no subscription. Guillotine is free and open source under AGPL-3.0.
Keyboard & mouse, first class.
On a Chromebook, tablet or desktop, Guillotine drives like a proper editor — shortcuts, mouse, and Ctrl-scroll zoom. On touch, long-press a clip to range-select across tracks.
From raw footage to a saved cut.
Import
Pull in video, audio and images straight from your device. Drop them on the timeline across as many tracks as you like.
Prompt & edit
Select a clip, type something like cut the silences, and let the AI mark the cuts — or do it by hand with split, trim, filters and keyframes.
Export
Render a real mp4 on-device. It lands in Movies/Guillotine in your gallery, ready to share.
The short answers.
Is it really free?
Do I need an account or internet connection?
Is my footage uploaded anywhere?
Where do my exports go?
Movies/Guillotine in your gallery, with progress shown as it renders.Does it work on a Chromebook or tablet?
Is there a desktop version?
How do I report a bug?
What can I control — formats, quality, aspect ratio?
Every platform, one editor.
Native apps for phone, tablet, Chromebook, and desktop. Shared editor core; each platform gets a real installer built on its own CI runner.
Desktop installers are currently unsigned. First launch:
macOS right-click → Open,
Windows SmartScreen → More info → Run anyway,
Linux sudo apt install ./guillotine_*.deb.
Cut to the chase.
Grab the latest build and start editing. Free, open source, and runs entirely on your device.