We’re excited to share that Stable Audio 3.0—Stability AI’s new family of music models built for artistic experimentation—is coming to ComfyUI. Trained on fully licensed data, these models bring variable-length generation, on-device-friendly small checkpoints, and stronger musicality for longer structure—so you can go from quick SFX to extended tracks inside the workflows you already use.
Model highlights
Licensed for commercial use — trained on fully licensed music data.
Flexible clip length — from quick SFX and short loops to longer tracks (up to about two minutes on Small, six minutes on Medium).
Lightweight, small models — run Small SFX and Small music on a CPU, no big GPU required.
Medium for longer music — fuller tracks with stronger structure when you have a GPU.
Available Models
Small-SFX: Sound effects and short ambiance, up to 2:00,
Small-Music: Short music and on-device-friendly loops, up to 2:00
Medium: Longer tracks with stronger structure and musicality, up to ~6:20
Small reaches two minutes (vs. 11s / 47s on Stable Audio Open). Medium goes beyond six minutes when you need length.
🤗 Comfy-Org/stable-audio-3 (for ComfyUI)
Examples
Music
A short idea for a complete piece (genre, instruments, mood, length).
Lo-fi hip-hop chill track with mellow electric piano, soft vinyl crackle, subtle synth pads, low-pass filtered drums, percussion loops, and soft plucked bass for a relaxed, dreamy vibe. BPM: 75. Length: 150 secondsSynthwave 80s retro track with arpeggiated synth leads, analog pads, electric bass, punchy electronic drums, gated reverb snares, and atmospheric FX for nostalgic and vibrant energy. BPM: 110. Length: 180 secondsInstrument
Single-instrument or small-group clips for production and scoring.
Guitar muted strum loop with tight rhythmic feel. BPM: 100. Length: 8 secondsPluck sequence loop with bright resonant tone. BPM: 128. Length: 10 seconds Sound effects and ambience
Impacts, textures, environments, and motion sounds for pictures and games.
Footsteps on gravel, steady walking pace, close perspective. Length: 8 secondsCar speeding past at high velocity, doppler effect, realistic whoosh. Length: 3 secondsOne-shot
Brief isolated sounds for drums, UI, transitions, and sample libraries.
Bass pluck with jazzy tone and resonant wooden body. Length: 3 secondsLatin drums, dynamic Latin drums and percussion ensemble featuring authentic rhythmic patterns. Length: 3 seconds
Get started
Update ComfyUI to v0.22.0 or go to Comfy Cloud
Go to the left sidebar → Template → Audio category → Choose Stable Audio 3.0 Template
For local users, please follow the note in the workflow to download the models and place them in the correct directory
Write a prompt, set the duration in seconds, then hit run.
As always, enjoy creating!
Yes, starting comfyUI without Sage Attention fixed the problem. Thank you.
The audio output is awful. I do not have sage attention enabled.