Skip to content
Audio

Geode ACE-Step 1.5 turbo: running music generation on smartphones

2 min readupdated

This model is a converted version of the fast ACE-Step 1.5 turbo music model (created by ACE Studio and StepFun), adapted to run directly on mobile phones via the Geode Android app. Its main purpose is generating songs on-device in just 8 steps without guidance, which is significantly faster than the standard sft version. The weights are optimized and quantized to int8, and the MIT license permits commercial use of the generated tracks.

Key features

  • Mobile optimization: weights are distributed to run within the Geode app on Android.
  • Fast generation: the DiT (Diffusion Transformer) turbo mode requires only 8 steps.
  • Flexible quantization: matrices are available in int8 (r8) formats for CPU, NPU, and GPU, alongside optional bf16/fp16 formats.
  • Commercial license: distributed under the MIT license, allowing generated songs to be used commercially.

What you can do

  • Music generation: create full tracks and songs directly on a compatible smartphone using the Geode app.
  • Hardware testing: test generative audio models running on mobile processors and NPUs.
  • Commercial projects: use the generated compositions for personal or commercial purposes thanks to the permissive license.

Example

Running the model requires the shared acestep-1.5/ package, which the Geode app downloads automatically if missing. The generation parameters use 8 sampling steps with a shift of 3.

Read moreDetails
Base model: ACE-Step/Ace-Step1.5
File size: the main package is 1.71 GiB, and the optional dit_turbo_f16.bin file is 2.62 GiB
Training details: the original checkpoint weights are distributed for the Geode engine, the per-step matrix is quantized to int8 (r8), and the rest is preserved in 16-bit
License: MIT
Discuss Telegram 0

Will it run?

Runs in F16

Fits without quantisation.

  • F16weights 2.6 GB · ready files3.4 GB
  • FP8 / INT8recommendedweights 1.3 GB1.9 GB
  • 4-bitweights 0.7 GB1.3 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · F16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # Maximilian228/Geode-ACE-Step-1.5-turbo
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download Maximilian228/Geode-ACE-Step-1.5-turbo --local-dir ~/models/Geode-ACE-Step-1.5-turbo
    # The model is in ~/models. How to run it is in the model card (link above).
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of Ace-Step1.5 (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 12All on Hugging Face
Merges 2All on Hugging Face

About the model

Maximilian228/Geode-ACE-Step-1.5-turbo
Author
Maximilian228
Category
Audio
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments

Audio

pingo-ptbr

pingo-ptbr