ACE-Step 1.5 🧠 Model Open source

What it is

ACE-Step is an open-source music generation model. ACE-Step 1.5, released on 2 April 2026, generates songs from 10 seconds to 10 minutes long, with vocals and lyrics in 50+ languages. It is published under the MIT licence. An XL variant with a 4B-parameter decoder improves quality for users with bigger GPUs.

Key features

• Songs from 10 seconds to 10 minutes, with vocals and lyrics
• 50+ lyric languages
• A full song in under 2 seconds on an A100, or under 10 seconds on an RTX 3090
• Runs in 6 GB or less of VRAM with quantisation and CPU offload. 12 GB+ is recommended for XL
• Official ComfyUI integration and a community ecosystem

Where you can use it

Clone the GitHub repo and run it locally with uv, or use the official ComfyUI nodes. We found no commercial creator app in our directory that documents ACE-Step by name.

Pricing and rights

Free under the MIT licence, including commercial use of the model. You pay only for your own GPU. The licence covers the software. You are still responsible for not imitating real artists or reproducing copyrighted lyrics in what you publish.

Who it is for

Musicians, ComfyUI users and developers who want an open, commercial-friendly song generator with vocals on their own hardware.

Verdict

ACE-Step 1.5 is the strongest permissively licensed song model with vocals, and it is fast. Vocal realism and mixing still trail Suno v6 and Lyria 3.5, and setup takes some technical comfort.