MiniMax Music 3 One-Click ComfyUI Windows Installer
Give it lyrics and a style description and it generates a complete song with vocals, instruments and full structure — up to 5 minutes in 32 kHz stereo. MiniMax Music 3 is one of the best local, open-weights alternatives to Suno, and this one-click installer sets up ComfyUI Portable, the required nodes, the workflow and all three models in a single double-click.
No subscription and no credit meter — it runs entirely on your own machine, and it's light enough for around 8 GB of VRAM.
Why this one installs cleanly: MiniMax Music 3 runs on native ComfyUI support rather than an exotic custom-node stack, so there's no fragile external dependency to break on update. The work the installer does is getting the correct ComfyUI Portable build in place and pulling the three model files — about 14.3 GB total — into the right folders.
What the installer does: sets up ComfyUI Windows Portable, installs the core support nodes (ComfyUI Manager, rgthree, Custom Scripts), preloads the song-generation workflow, then pauses to ask before downloading the models. Git, 7-Zip and the correct PyTorch build are handled automatically.
A small footprint for a full song generator — and unlike a hosted service, every track you make stays on your machine.
What you get
A one-click Windows installer that sets up ComfyUI Portable, the support nodes, the song-generation workflow, and all three MiniMax Music 3 models.
Model files (downloaded automatically into the right folders — three files, ~14.3 GB total):
minimax_music3_dit_fp16 (diffusion) — 4.91 GB — placed in unet/
minimax_music3_text_encoder_pruned_int8_convrot — 9.2 GB — placed in clip/
minimax_music3_dav (VAE) — 217 MB — placed in vae/The FP16 DiT is the recommended build for quality; the text encoder is the largest single file.
Support nodes included: ComfyUI Manager, rgthree, Custom Scripts. MiniMax Music 3 itself uses native ComfyUI nodes, so there is no external node stack to break.
System requirements
OS: Windows 10 or 11, 64-bit.
GPU: NVIDIA RTX 30XX / 40XX / 50XX. Around 8 GB VRAM should be enough for the FP16 build; more headroom helps with longer tracks.
Disk: at least 25 GB free (models + ComfyUI + working space).
Handled for you: Git, 7-Zip and the correct PyTorch build are installed automatically.
Getting started
1. Put the install .bat in its own dedicated folder and double-click. Use a simple local path like C:\MiniMaxMusic\ — avoid Desktop, Documents, and cloud-synced folders.
2. The installer pauses before downloading models — press Y to download or N to skip. The window waits for your input.
3. Load the workflow and confirm all three models are selected in the loaders.
4. Write a good caption — this is where quality lives. Structure it in three parts: global metadata (genre, BPM, key, mood, production), vocal details (gender, tone, delivery), and arrangement (how sections evolve). Tag your lyrics with [Verse] and [Chorus] on their own lines. The easiest way to get this right is to have an LLM format your song idea into that structure while keeping your lyrics intact.
5. Keep first generations short to dial in your caption and vocal style before committing to a full-length track.
Troubleshooting
Install fails or behaves oddly: check the install path first. A cloud-synced folder (OneDrive, Dropbox) or a deep path with spaces is the most common cause. Move to C:\MiniMaxMusic\ and re-run.
Output doesn't match what you asked for: this is almost always the caption, not the model. Rewrite it in the three-part structure above — vague captions produce vague songs.
Generation is slow: time scales with song length and GPU. Test your caption on a short clip before generating a full track.
Out of memory: shorten the track. Length is the main driver of memory use on the FP16 build.
A model isn't showing in a loader: confirm you pressed Y at the download prompt rather than skipping, and check the three files landed in unet/, clip/ and vae/.
Licence & credit
MiniMax Music 3 is by MiniMax. This product is the installer and workflow packaging — it automates the setup of their model and the native ComfyUI workflow into a one-click process.
Licence note: MiniMax Music 3 uses MiniMax's own community licence, not a fully open one. It is free for personal use. If you monetise your output, you are asked to credit "MiniMax-Music3" and to disclose the audio as machine-generated. Full terms are on the model's Hugging Face page — read them before commercial use.