Model reference · open weights
ACE-Step is an open-weight audio or speech model from ACE-Step. ACE-Step-v1-3.5B (BF16) weighs 8.3 GB; the smallest configuration that runs it is RTX 3060 12 GB.
Summary of the ACE-Step/ACE-Step-v1-3.5B model card, 2026-10-01
What it is
| Released by | ACE-Step |
|---|---|
| Released | 2025-04-28 |
| VRAM | 8.3 GB for the weights |
What it runs on
| Card | One stream | Memory |
|---|---|---|
| RTX 3060 12 GB | fits | 11.6 GB |
| RTX 4060 Ti 16 GB | fits | 15.4 GB |
| RTX 3090 24 GB | fits | 23.4 GB |
| RTX 4090 24 GB | fits | 23.4 GB |
| RTX 5090 32 GB | fits | 31.0 GB |
| L40S 48 GB | fits | 44.0 GB |
| A100 80 GB | fits | 78.2 GB |
| H100 80 GB | fits | 78.1 GB |
| RTX PRO 6000 Blackwell 96 GB | fits | 93.8 GB |
| DGX Spark (GB10) 128 GB unified | fits | 107 GB |
| H200 141 GB | fits | 138 GB |
| B200 180 GB | fits | 176 GB |
From the model card
ACE-Step is a novel open-source foundation model for music generation that overcomes key limitations of existing approaches through a holistic architectural design. It integrates diffusion-based generation with Sana's Deep Compression AutoEncoder (DCAE) and a lightweight linear transformer, achieving state-of-the-art performance in generation speed, musical coherence, and controllability.
Key Features:
ACE-Step can be used for:
The model serves as a foundation for:
The model should not be used for:
see: https://github.com/ace-step/ACE-Step
| Device | 27 Steps | 60 Steps |
|---|---|---|
| NVIDIA A100 | 27.27x | 12.27x |
| RTX 4090 | 34.48x | 15.63x |
| RTX 3090 | 12.76x | 6.48x |
| M2 Max | 2.27x | 1.03x |
RTF (Real-Time Factor) shown - higher values indicate faster generation
Users should:
Developed by: ACE Studio and StepFun Model type: Diffusion-based music generation with transformer conditioning License: Apache 2.0 Resources:
@misc{gong2025acestep,
title={ACE-Step: A Step Towards Music Generation Foundation Model},
author={Junmin Gong, Wenxiao Zhao, Sen Wang, Shengyuan Xu, Jing Guo},
howpublished={\url{https://github.com/ace-step/ACE-Step}},
year={2025},
note={GitHub repository}
}
This project is co-led by ACE Studio and StepFun.
Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.
Running it yourself
Rent a machine by the hour. ComfyUI is installed on it. Open ComfyUI through the tunnel, then Templates → ACE Step v1 Text to Song: it loads ComfyUI's own build of this model, the files above.
# on your rented machine (the ssh line is on its page in the console)
# get REPO FILE FOLDER: one file into /workspace/models/FOLDER, where ComfyUI loads it from
get() { hf download "$1" "$2" --local-dir /workspace/hf-files && mkdir -p "/workspace/models/$3" && mv "/workspace/hf-files/$2" "/workspace/models/$3/$4"; }
# the files of the template “ACE Step v1 Text to Song” (7.2 GB)
get Comfy-Org/ACE-Step_ComfyUI_repackaged all_in_one/ace_step_v1_3.5b.safetensors checkpoints
start-comfyui# on your computer, in a second terminal: ComfyUI in your browser at http://localhost:8188
# HOST and PORT are your machine's, from its page in the console
ssh -L 8188:localhost:8188 dev@HOST -p PORT