Mondo Music Video turns a song into a beat-cut music video with dancing and instrument-playing actors, rendered locally or through APIs
Mondo Music Video is a free, Apache-2.0 studio that turns a song into a music video: it detects beats, downbeats, energy and sections, plans timed shots, has an LLM write the prompts and renders them locally in ComfyUI or through hosted video APIs. Version 1.12.4, released on 7 November 2026, adds optional storyboard text cleanup and eight more thematic Worlds.
What it does
A reviewer called Laya checks the prompts before rendering, then Mondo trims every passage to its planned frames, joins the picture and puts your original song back. Actors from reference images can dance to the beat, drums, bass or vocals, lip-sync and play instruments, and you can add storyboards, lyric captions, credits and idents; a queue (since 1.12.0) renders productions one after another.
Worth knowing
You bring your own models and API keys: model weights and commercial software are not bundled. Local work runs without confirmation, while paid API work only starts after you accept its cost estimate and never retries itself. A coding agent can install it from the repository's guide, which checks your hardware and asks before large downloads; the local MiniMax H3 path runs through the official Comfy SDK proxy, so the H3 weights licence applies.