AI video, scripted
The best AI models
for makingany kind
of video.
mograf is a declarative pipeline for AI video. Write a timeline, run it from the CLI, render with the strongest model for the job. Share the recipe so anyone can re-run it.
Best-of-breed, model-agnostic
Video
Seedance
1.1 Pro · 4-15s clips
Image
Nano Banana
Gemini 3.1 Flash · Pro
Image
GPT Image 2
OpenAI · refs + edits
Voice
Gemini 3.1 TTS
200+ inline audio tags
Voice
ElevenLabs
forced alignment, clones
Transcribe
Deepgram
low-latency captions
Workflows for every kind of video
Each workflow is a recipe: README, a main rendered example, the timeline that produced it, plus any reference packs it needs. Pushed by the community, runnable in one command.
narration
Narrated explainer
TTS + AI visuals, beat-by-beat, aligned to your script.
news
News-show clip
Recurring anchor, lower-thirds, recurring set, multi-clip consistency.
music
Music video
Cuts driven by beat detection; per-beat prompts; reusable style.
style-rip
Style-rip
Sample any reference clip → reusable style pack you can apply.
Push a workflow in one command
Sign in with GitHub, link the CLI on each machine you use, and anything you generate locally is one command away from a public, re-runnable URL.
# install the CLI globally (uv recommended; pipx also works)
uv tool install mograf
# GitHub device flow + register this machine
mograf login
# publish a pipeline run as a shareable workflow
mograf publish runs/your-latest-run