★ Reading this for free? Get 20 structured AI courses + per-chapter AI tutor — the first chapter of every course free, no card.Start free in 30 seconds
Audio

MiniMax Music 3 on 8GB VRAM: Full Songs With Vocals

September 13, 2026
13 min read
LocalAimaster Research Team

Want to go deeper than this article?

Free account unlocks the first chapter of all 25 courses — RAG, agents, MCP, voice AI, MLOps, real GitHub repos.

📚AI Learning Path

Got the hardware sorted? Now build on it. You know what to buy — the courses show you what to actually run, fine-tune, and ship on it. First chapter free, no card.

Start free
Or own it for life — Lifetime $149, pay once

Short answer: yes, MiniMax Music 3 runs on an 8GB card — but the 8GB path is a streaming trick, not a fit. MiniMax's own model card says full precision "fits under 24GB", drops to ~22GB with CPU offload, and only reaches 8GB by streaming the language model layer by layer. The thing eating your VRAM is not the diffusion model (2.5GB at int8) — it is the 18.47GB text encoder. Output is 32kHz 16-bit stereo, up to five minutes, with sung vocals.

That is the whole page in a paragraph. What follows is the file-by-file detail: exactly which weights to download for your card, where the ComfyUI route differs from the diffusers route, and the two clauses in the licence you need to read before you put a single track on Spotify.

One disclosure up front, because it is load-bearing: we have not yet timed generation on an 8GB card ourselves. The 8GB claim on this page is MiniMax's, quoted from the model card. Every file size, licence term, and architecture number below is read straight off the official repositories. Where a number is a vendor claim, it is labelled as one.


What It Is

MiniMax Music 3 is an open-weights text-to-music model that writes and sings complete songs — lyrics, vocals, arrangement — from two text inputs. It landed on Hugging Face on 7 August 2026 and has been updated since (last revision 14 August). Reading the Hugging Face counters on 18 August 2026, the main repo showed roughly 10,400 downloads and 920 likes, with a community GGUF conversion at about 2,700 downloads. Those are 30-day rolling figures and they move daily — treat them as a snapshot, not a scoreboard. For a music model two weeks old, it is fast uptake.

The architecture, per the model card:

ComponentSizeJob
Global LLM8B (initialised from Qwen3-8B)Long-range structure; predicts the first RVQ codebook frame by frame
Local LLM0.6BFrame-level acoustic detail; the remaining codebooks
Flow Matching2.4BContinuous hidden-state synthesis
Flow-VAE decoder123MFinal 32kHz stereo audio

The tokeniser uses eight layers of residual vector quantisation — a 16,384-entry semantic codebook plus seven 1,024-entry acoustic codebooks. At inference the synthesis path fuses the LLM hidden states directly and skips the discrete tokeniser decoder entirely.

What that buys you in practice is the thing every previous open music model failed at: a song that still remembers its own chorus four minutes in. Intro, verse, pre-chorus, chorus, bridge, instrumental, outro — with a consistent vocal identity across the whole thing.


Reading articles is good. Building is better.

Free account = the first chapter of all 25 courses, with a per-chapter AI tutor. No card.

Will It Fit? Per-VRAM Table

Pick your row, download those three files, ignore everything else. These sizes are read directly from the Comfy-Org repack and the community GGUF repo file listings — they are exact, not estimates.

Your GPUDiffusion modelText encoderVAEOn-diskReality check
8GB (3060 Ti, 4060, 3070)GGUF Q4_K_M — 1.49GBpruned int8 — 9.20GB0.217GB~10.9GBWorks only with aggressive offload; the encoder lives in system RAM. Slow.
12GB (3060 12GB, 4070, 5070)int8_convrot — 2.50GBpruned int8 — 9.20GB0.217GB~11.9GBThe sensible floor. Still offloading, but far less thrash.
16GB (4060 Ti 16GB, 4080, 5070 Ti)fp16 — 4.91GBpruned int8 — 9.20GB0.217GB~14.3GBFirst tier where the whole stack is nearly resident.
24GB (3090, 4090, 5090 has more)fp16 — 4.91GBpruned bf16 — 16.71GB0.217GB~21.8GBMatches the card's "full precision fits under 24GB".
32GB+ (5090, RTX Pro, A6000)fp32 — 9.83GBbf16 — 18.47GB0.217GB~28.5GBNo compromises, no offload, longest songs.
Apple SiliconNo supported path. The card lists "Inference requires CUDA".

The community GGUF ladder for the diffusion transformer, if you want to trade fidelity for headroom (sizes from the repo listing):

QuantSizeRepo's own note
F164.98GBUncompressed reference baseline
Q8_02.70GBNear-lossless (the repo's recommended pick)
Q6_K2.12GBExcellent fidelity
Q5_K_M1.79GBBalance of speed and footprint
Q4_K_M1.49GBFlagged for 8GB–10GB GPUs
Q3_K_M1.16GB"May introduce audible softening"

System RAM matters more here than on any LLM you have run. Everything that does not fit in VRAM sits in RAM and gets streamed across PCIe every step. Our working rule for the 8GB and 12GB rows: budget 32GB of system RAM, and do not attempt it on 16GB while a browser is open. That is our guidance from how offloading behaves generally, not a number from MiniMax.

If your card is the constraint rather than your patience, the hardware hub and our GPU ranking for local AI cover what a step up actually buys. A used 24GB card — see the RTX 3090 guide — puts you straight into the "no compromises" row for less than a new mid-tier GPU.


The Text Encoder Trap

The single most useful thing to understand: MiniMax Music 3's "text encoder" is an 8B language model, and it is 80% of your memory problem.

Look at the numbers side by side. Comfy-Org ships three text encoders:

Text encoder fileSize
minimax_music3_text_encoder_bf16.safetensors18.47GB
minimax_music3_text_encoder_pruned_bf16.safetensors16.71GB
minimax_music3_text_encoder_pruned_int8_convrot.safetensors9.20GB

And three diffusion models:

Diffusion model fileSize
minimax_music3_dit_fp32.safetensors9.83GB
minimax_music3_dit_fp16.safetensors4.91GB
minimax_music3_dit_int8_convrot.safetensors2.50GB

The 18.47GB "text encoder" lines up almost exactly with the 18.48GB qwen_7B directory in the original repo. It is the Qwen3-8B-derived Global LLM wearing a ComfyUI hat.

This is why the GGUF quants, useful as they are, do not solve the 8GB problem on their own. Going from F16 (4.98GB) to Q3_K_M (1.16GB) saves you 3.8GB. Going from the bf16 encoder to the pruned int8 encoder saves you 9.3GB. Get the encoder choice right first; the DiT quant is a rounding error next to it.

Practical consequence for the 8GB row: even the smallest encoder is 9.20GB, which is bigger than your entire card. Nothing you do makes it resident. The model runs because ComfyUI and diffusers page pieces of it in and out — which is exactly what MiniMax means by "streaming the language model layer by layer".


Setup: ComfyUI

ComfyUI has native support and an official template — this is the easiest route, and the only one with a low-VRAM story that does not involve writing Python.

Per the ComfyUI docs, the flow is: update ComfyUI to the latest version, open Template Library → Audio → MiniMax Music 3, and let the pop-up fetch the models. If the nodes are missing, your ComfyUI is too old.

If you would rather pull the weights yourself, these are the exact commands from the repos:

# Diffusion model (int8 — the 12GB pick)
huggingface-cli download Comfy-Org/MiniMax-Music-3 \
  diffusion_models/minimax_music3_dit_int8_convrot.safetensors \
  --local-dir ComfyUI/models/diffusion_models

# Text encoder (pruned int8 — the low-VRAM pick)
huggingface-cli download Comfy-Org/MiniMax-Music-3 \
  text_encoders/minimax_music3_text_encoder_pruned_int8_convrot.safetensors \
  --local-dir ComfyUI/models/text_encoders

# VAE
huggingface-cli download Comfy-Org/MiniMax-Music-3 \
  vae/minimax_music3_dav.safetensors \
  --local-dir ComfyUI/models/vae

For the GGUF route you also need the ComfyUI-GGUF custom node, and you load the model through Unet Loader (GGUF) instead of the standard loader:

huggingface-cli download Abiray/MiniMax-Music3-GGUF \
  MiniMax-Music3-Q4_K_M.gguf \
  --local-dir ComfyUI/models/diffusion_models

Two workflow controls are worth knowing before your first run, both documented by ComfyUI:

  • max_duration — target length in seconds. The template defaults to 60; the model supports up to about 300. Longer costs more time and more VRAM.
  • tiled_decode — decodes the audio VAE in overlapping tiles to cut VRAM. ComfyUI's own note: helpful for long songs on low-VRAM GPUs, slightly slower, small risk of seams at tile boundaries. Turn it on for the 8GB and 12GB rows. Turn it off above that.

New to ComfyUI generally? Start with our ComfyUI complete guide. If it loads and then dies mid-generation, that is almost always a memory-management problem rather than a MiniMax problem — our ComfyUI out-of-memory guide covers the usual causes.


Own it instead of renting it

Run this on your own machine and stop paying every month

Pay once and keep it. No renewal, no per-token bill, and nothing you feed it ever leaves your hardware.

Setup: Diffusers and SGLang

If you want the 8GB path exactly as MiniMax describes it, you need the diffusers route — the group-offloading hook is what makes the claim true.

At the time of writing the diffusers integration is still an unmerged PR, so the card pins a specific commit:

pip install git+https://github.com/huggingface/diffusers@dafe3733fcfdbf3c48915fe77be3aef65b5d6a2d \
  transformers accelerate soundfile

The card's own low-VRAM snippet, which is the load-bearing part:

import torch
from diffusers import ComponentsManager, ModularPipeline
from diffusers.hooks import apply_group_offloading

manager = ComponentsManager()
manager.enable_auto_cpu_offload(device="cuda")
pipe = ModularPipeline.from_pretrained("MiniMaxAI/MiniMax-Music3", components_manager=manager)
pipe.load_components(dtype=torch.bfloat16)

# Only needed below ~22 GB of VRAM — slower, but fits in 8 GB.
apply_group_offloading(
    pipe.language_model, onload_device=torch.device("cuda"),
    offload_type="leaf_level", use_stream=True
)

Note the comment MiniMax left in their own example: "slower, but fits in 8 GB." They are not overselling it.

There is also an SGLang-Omni server path, which is the one to use if you are generating in volume rather than experimenting:

hf download MiniMaxAI/MiniMax-Music3 --local-dir /path/to/minimax_ttm
sgl-omni serve --model-path MiniMaxAI/MiniMax-Music3 --port 8000

It exposes an OpenAI-shaped speech endpoint — lyrics go in input, the music description goes in instructions:

curl http://127.0.0.1:8000/v1/audio/speech \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "MiniMaxAI/MiniMax-Music3",
    "input": "[Verse]\nMorning light filtering through the pine\n[Chorus]\nSoftly the world begins to breathe",
    "instructions": "A warm acoustic pop song with intimate female vocals, fingerpicked guitar, soft piano, and a gradual emotional build into a wide final chorus.",
    "response_format": "wav",
    "seed": 7,
    "max_new_tokens": 750,
    "stream": false
  }' --output minimax_music3.wav

max_new_tokens counts audio frames at 25 frames per second, so 750 frames is 30 seconds. A five-minute song is 7,500 — and the hard ceiling is 9,000 frames.

Be aware of the download. hf download on the original repo pulls roughly 57GB (we summed the file listing: 18.5GB Qwen encoder, 17.2GB language model, 9.8GB Flow-VAE, 9.7GB transformer, 1.3GB RVQ depth decoder, plus the rest). The ComfyUI repack exists precisely so you do not have to do that.


Prompting: Caption + Lyrics

Two inputs, and the model is far more responsive to the caption than most people expect. MiniMax recommends a structured caption in three named sections:

  • Global Metadata — genre, subgenre, BPM, key, scale, emotional progression, listening scenario, production profile
  • Vocal Details — vocal gender, timbre, performance style, harmony, backing vocals, vocal effects
  • Arrangement — primary and secondary instruments, section-level instrument evolution, groove, bass, percussion, textures, spatial effects

Lyrics carry the structure through section tags on their own lines: [Intro], [Verse], [Pre-Chorus], [Chorus], [Post-Chorus], [Bridge], [Instrumental], [Solo], [Outro].

The example from the official repo, verbatim:

Genre: acoustic pop. BPM: 96. Key: C major. Warm and intimate, building gently
into the chorus. Vocals: soft female lead, close and breathy, light stacked
harmonies in the chorus. Arrangement: fingerpicked guitar and soft piano;
brushed drums and upright bass enter in the chorus.

MiniMax also ship a music-caption-rewriter skill that expands a one-line idea into the full structured caption:

npx skills add MiniMax-AI/MiniMax-Music3 --skill music-caption-rewriter

One honest caveat from the model card's own limitations list: section tags and descriptions provide generative control rather than strict symbolic guarantees. Ask for 96 BPM in C major and you will usually get something near it, not a metronome-accurate render. Plan on re-rolling seeds.


The Licence: Read This Before You Monetise

MiniMax Music 3 is not open source in the OSI sense. It ships under the MiniMax-Music3 Community Licence, and commercial use carries two specific obligations. Quoting the LICENSE file directly:

  1. Attribution is mandatory and visible. "You shall prominently display 'MiniMax-Music3' on the user interface of commercial product or service that uses the Software." Not a footnote in your terms — on the interface.
  2. A $20M revenue ceiling. Above "20 million US dollars" in aggregate yearly revenue across you and your affiliates, you need "separate, prior written authorization from MiniMax" via api@minimax.io.
  3. Safeguards if you host it for others. If you offer a product or service that lets third parties generate outputs, you must implement, maintain, test and periodically review "reasonable and proportionate technical and organizational safeguards" against infringing outputs — and not weaken them.
  4. An Acceptable Use Policy MiniMax can revise. Exhibit A runs to 19 clauses. Two that bite ordinary creators: no undisclosed machine-generated content in public environments, and no impersonation without consent.

For most independent producers, clauses 1 and 4 are the whole story, and clause 2 will never apply. That is a genuinely usable deal — considerably better than the research-only licences elsewhere in this space. But it is a licence with conditions, not a permissive one, and "prominently display MiniMax-Music3" is a real product-design constraint if you were planning to white-label a song generator.

One correction worth flagging. The community GGUF repository tags itself license: apache-2.0 and its README says it "inherits the Apache-2.0 License from the original release". That is incorrect. The original release carries the MiniMax Community Licence — the LICENSE file is right there in the repo. Do not rely on a third-party quantiser's frontmatter for your legal position. (Underlying components have their own permissive licences — Qwen3-8B is Apache-2.0, the DiT derives from Stable Audio under MIT, the VAE from DAC under MIT — but the composed model is governed by MiniMax's terms.)

If licence terms are the deciding factor for you, the contrast with the voice side of this stack is stark: VoxCPM2 is straight Apache-2.0, no attribution clause, no revenue ceiling.


Honest Limitations

Straight from the model card's own limitations list, plus what follows from the file sizes:

  • CUDA required. No Apple Silicon, no ROCm path documented. This is the biggest exclusion.
  • Non-streaming generation only. You wait for the whole song; there is no play-as-it-renders.
  • 5,000-token prompt cap, and a 9,000 acoustic frame ceiling (360 seconds).
  • Control is soft. Tempo, key, instrumentation, lyrics and structure "may not always match every requested detail exactly".
  • The 8GB path is a fallback, not a target. MiniMax's own code comment calls it slower. We have not measured how much slower.
  • Nothing here is a mixing engineer. You get a finished-sounding stereo file, not stems.

What we could not verify and are not going to guess at: generation time per minute of audio on any specific card, real peak VRAM under load (as opposed to on-disk file sizes), and quality deltas between the GGUF quant levels. When we have run it on our own hardware, this page gets the numbers.


vs ACE-Step and Suno

Aimed squarely at long-form coherence and vocals, where the open incumbents are weakest — but ACE-Step is still the far bigger download, and nobody has published a blind listening test either way. Against Suno you give up polish and convenience; you get weights you keep.

Against ACE-Step, the incumbent open text-to-music model: be clear that ACE-Step is still the far bigger download. On 18 August 2026 the Hugging Face 30-day counters put ACE-Step 1.5 at roughly 57,000 on the main repo, with its ComfyUI repack and community GGUFs adding several hundred thousand more, against MiniMax Music 3's ~10,400. MiniMax is the newcomer, not the incumbent. What it is aimed at is exactly ACE-Step's weak spot — five-minute structural coherence with a stable vocal identity, rather than a good-sounding 30-second loop. The trade is size: ACE-Step runs comfortably where MiniMax Music 3 needs offloading. We have not run a blind listening comparison between the two, and we are not going to imply one.

Against Suno: Suno wins on polish, iteration speed, and not requiring a GPU. MiniMax Music 3 wins on the three things that make people search for a local alternative in the first place — no per-song credit meter, no upload of your lyrics to someone's server, and weights you keep. If you generate five songs a month, Suno is cheaper and better. If you generate five hundred, or your lyrics are under NDA, or you want to fine-tune, this is one of the few open models where that trade is genuinely close.

For the broader landscape of local music tooling, see our local AI music and DJ tools roundup and the music generation tutorial — both predate this model, and MiniMax Music 3 is now the answer they were waiting for.


Verdict

  1. The 8GB headline is true but load-bearing on a caveat. It fits by streaming an 8B language model layer-by-layer from system RAM. It will finish. We cannot yet tell you how long you will wait.
  2. 12GB is the honest floor, 16GB is comfortable, 24GB matches the card's "full precision" tier. Pair any of them with 32GB of system RAM.
  3. Choose the text encoder first. Pruned int8 (9.20GB) versus bf16 (18.47GB) is a 9GB decision. The DiT quant is worth 2-4GB.
  4. Use ComfyUI unless you have a reason not to. Native template, tiled decode, ~12GB of downloads instead of 57GB.
  5. Read the LICENSE, not the GGUF repo's frontmatter. Community licence, visible attribution required, $20M revenue gate. Usable for almost everyone — but it is a condition, not a freedom.

"Make a full song with vocals, on my own machine, and sell it" is a real sentence now — though be honest that MiniMax Music 3 is not the only model it applies to. ACE-Step 1.5 carries a plain MIT licence, which is strictly less restrictive than MiniMax's community terms. What MiniMax brings is the architecture built for five-minute structure with one consistent voice. The hardware bill is steeper than the headline suggests, and the licence has strings. Both are still a much better deal than a credit meter.


FAQ

🎯
AI Learning Path

Got the hardware sorted? Now build on it.

You know what to buy — the courses show you what to actually run, fine-tune, and ship on it. First chapter free, no card.

Or own it for life — Lifetime $149 $599, pay once
Once your hardware is sorted

Decide before you spend a thousand pounds

The AI Hardware course sizes your build properly — VRAM ladder, real bottlenecks, budget builds — and Pick the Right Model tells you what to run on it.

$149 once unlocks everything, forever — about $0.27/chapter for life. Prefer to spread it out? Pro is $79/year (saves 27%) or $8.99/month.
Secure checkout by Lemon Squeezy — your card never touches this siteInstant access the moment you payFirst chapter of every course is free — try before you buy

Liked this? 25 full AI courses are waiting.

From fundamentals to RAG, agents, MCP servers, voice AI, and production deployment with real GitHub repos. First chapter free, every course.

Reading now
Join the discussion

LocalAimaster Research Team

Creator of Local AI Master. I've built datasets with over 77,000 examples and trained AI models from scratch. Now I help people achieve AI independence through local AI mastery.

Build Real AI on Your Machine

RAG, agents, NLP, vision, and MLOps - chapters across 25 courses that take you from reading about AI to building AI.

Want structured AI education?

25 courses, 519+ chapters, from $9. Understand AI, don't just use it.

AI Learning Path
More on Local AI Hardware
See the full AI Hardware Guide 2026 guide.

Comments (0)

No comments yet. Be the first to share your thoughts!

Can MiniMax Music 3 really run on an 8GB GPU?

Per the official model card, yes — but by streaming, not by fitting. The card states the full-precision model "fits under 24GB of VRAM", drops to "~22 GB" with automatic CPU offloading, and that "streaming the language model layer by layer makes it fit even 8 GB video cards" using diffusers group offloading at leaf level. That is a MiniMax claim, not a benchmark we have run. Streaming an 8B language model layer-by-layer over PCIe is inherently slow, so treat 8GB as "it will complete" rather than "it will be pleasant". If you have 16GB or more, do not use the streaming path.

How long can a song be, and what format does it output?

32kHz, 16-bit stereo WAV, up to five minutes native. The model card puts the hard ceiling at 9,000 acoustic frames at 25 frames per second — 360 seconds of audio — and caps the tokenised text prompt at 5,000 tokens. The ComfyUI template defaults to 60 seconds and the docs describe support "up to about 300 seconds / 5 minutes". Longer songs cost more time and more VRAM.

What is the total download size?

Two very different answers. Cloning the original MiniMaxAI/MiniMax-Music3 repo pulls about 57GB of weights (we summed the file listing: 18.5GB Qwen encoder, 17.2GB language model, 9.8GB Flow-VAE, 9.7GB transformer, plus decoders). The ComfyUI route is much smaller because Comfy-Org repacked it: the int8 diffusion model is 2.50GB, the pruned int8 text encoder is 9.20GB, and the VAE is 0.217GB — under 12GB total.

Can I sell music made with MiniMax Music 3?

Under the MiniMax-Music3 Community Licence, commercial use is permitted with two named conditions: you must "prominently display MiniMax-Music3 on the user interface" of any commercial product or service using the software, and if your aggregate yearly revenue exceeds 20 million US dollars you must get prior written authorisation from MiniMax. It is not an OSI-approved licence and it carries an Acceptable Use Policy MiniMax can revise. Note that the community GGUF repo labels itself Apache-2.0 — that label is wrong; the weights are governed by the LICENSE file in the original repo.

Does it work on a Mac?

Not on the supported path. The model card lists "Inference requires CUDA" as its first limitation, so Apple Silicon has no official route today — no MPS backend, no Metal build. Mac users with big unified memory pools are, ironically, worse off here than a $250 8GB Nvidia card. If a Metal path lands we will update this page.

Do I need the GGUF quants?

Only if you are on ComfyUI and short on VRAM, and understand what they cover. The community GGUFs quantise the diffusion transformer only — F16 4.98GB down to Q3_K_M 1.16GB — and still require the separate text encoder and VAE from the Comfy-Org repack. Since the text encoder is the 9-18GB component, GGUF-ing the DiT saves you a couple of gigabytes, not the whole problem.

Ready to Go Beyond Tutorials?

20 structured courses with hands-on chapters - build RAG chatbots, AI agents, and ML pipelines on your own hardware.

Was this helpful?

📅 Published: September 13, 2026🔄 Last Updated: September 13, 2026✓ Manually Reviewed
LM

Written by the Local AI Master Team

The team behind Local AI Master

We build Local AI Master around practical, testable local AI workflows: model selection, hardware planning, RAG systems, agents, and MLOps. The goal is to turn scattered tutorials into a structured learning path you can follow on your own hardware.

✓ Local AI Curriculum✓ Hands-On Projects✓ Open Source Contributor
📚
Free · no account required

Grab the AI Starter Kit — career roadmap, cheat sheet, setup guide

No spam. Unsubscribe with one click.

🎯
AI Learning Path

Go from reading about AI to building with AI

20 structured courses. Hands-on projects. Runs on your machine. Start free.

Or own it for life — Lifetime $149 $599, pay once
Free Tools & Calculators