ACE Music v1 Tops the Leaderboard: Create Studio-Quality Songs From Lyrics or Hums
ACE Music v1 just hit the top of Artificial Analysis' vocal music leaderboard, scoring 1160 Elo points against Suno v6's 1144. Launched November 11, it's the first new model to reach #1 in blind listening tests without being Suno. That alone would be news. But what's drawing creator attention right now is how ACE gives you control at every step.
What's Actually Shipping
ACE Music v1 is a text-to-song AI model from ACE Studio, launched November 11, 2026. You can try it free at acemusic.ai: 20 credits a day gets you up to 10 songs, but with no downloads or commercial use on the free tier.
Its new model, ACE Music v1, ranks #1 on Artificial Analysis's blind-listening vocal leaderboard, just ahead of Suno v6. The company also released a sibling model that first writes a score in ABC notation (melody, rhythm and chords) and then generates audio from it.
What Creators Are Doing With It
The songwriting workflow is the standout. Its "Idea" feature can develop a basic concept into lyrics and a structured composition, identifying individual song sections and describing how they should be arranged. Users can then refine the material rather than repeatedly generating completely new tracks.
Once you have a track, the tools stack up: Users can edit melodies and chords, regenerate performances, export MIDI files and separate vocals from instruments for further production. For creators working in DAWs, this means you generate a song, export it as MIDI to edit the arrangement in your usual workflow, then split stems to remix individual parts.
Melody to song: play notes on the browser keyboard, upload MIDI or hum into the mic, and ACE Music v1 builds a full track around it. That reverse workflow—starting from a hum or keyboard sketch—appeals to musicians who already have a melody in their head.
Open-Source Alternative Already Running Locally
If you want to experiment without paying or waiting for queues, ACE-Step 1.5 is an open-source music model that can generate a full song in about 2 seconds on an A100, runs locally on a typical PC (around 4GB VRAM), and beats leading commercial and open-source models on common evaluation scores. ComfyUI provides the most user-friendly way to use ACE-Step 1.5, with a visual node-based workflow that makes every feature accessible.
The ACE-Step v1 music foundation model in ComfyUI supports three powerful ways to create and edit music, with nodes for model loading, lyrics (supporting 19 languages), style and instrument control, and audio transformation.
What's Good, What's Not
The leaderboard win is real: In Decrypt's hands-on test, ACE Music v1 produced stronger melodies, more natural voices and cleaner endings, while Suno still handled some niche genres better. The song endings are noticeably better than competitors—no weird artifacts or abrupt stops. The scoring and MIDI features let you keep iterating instead of hitting "generate" over and over.
Limitations: ACE Music v1 runs only in the ACE Music web app for now, so there's no API or self-hosting the flagship model yet. The developers are pursuing a parallel open-source strategy and plan to open-source ACE Music v1 Flash in late 2026. If you need an API today, the open ACE-Step 1.5 is your route.
How to Try It
Log in to ACE Music and choose ACE Music v1 on the Create page. If you're new to ACE Music, finish the initial setup first. Free daily credits are enough to test the model. Downloads and commercial use start on the Creator plan at $7.90 a month billed yearly.
If you're already in ComfyUI, the ACE-Step 1.5 route lets you keep everything local and free. Install the ACE-Step custom nodes for ComfyUI, download the ACE-Step 1.5 model weights from Hugging Face or the official GitHub, place the model files in your ComfyUI models/ directory, and launch ComfyUI and load the ACE-Step workflow.