YOU ARE VIEWING ONE ITEM FROM THE AICRIER FEED

ACE-Step 1.5 XL targets higher quality

AICrier tracks AI developer news across Product Hunt, GitHub, Hacker News, YouTube, X, arXiv, and more. This page keeps the article you opened front and center while giving you a path into the live feed.

// WHAT AICRIER DOES

7+

TRACKED FEEDS

24/7

SCRAPED FEED

Short summaries, external links, screenshots, relevance scoring, tags, and featured picks for AI builders.

ACE-Step 1.5 XL targets higher quality
OPEN LINK ↗
// 55d agoMODEL RELEASE

ACE-Step 1.5 XL targets higher quality

ACE Studio and StepFun say ACE-Step 1.5 XL is coming, adding a 4B-parameter DiT decoder aimed at higher audio quality. The open-source music model still targets local use, with variants tuned for different quality and performance tradeoffs across Mac, AMD, Intel, and CUDA hardware.

// ANALYSIS

This is the kind of update that makes open-source music generation feel less like a novelty and more like a real alternative to closed tools. The catch is that the quality jump comes with a clear hardware split: the base model stays consumer-friendly, while XL pushes users toward much fatter VRAM budgets.

  • XL adds a 4B-parameter DiT decoder, with 12GB VRAM needed at minimum and 20GB recommended
  • The base ACE-Step 1.5 model still runs under 4GB VRAM, so the project keeps a strong local-first story
  • Broad platform support matters here: Mac, AMD ROCm, Intel XPU, CPU, and CUDA all get a path
  • The feature set goes beyond plain text-to-music into covers, repainting, stem separation, and vocal-to-BGM conversion
  • If the benchmark claims hold up, this raises the bar for open music models in the Suno/Udio orbit
// TAGS
audio-genopen-sourceinferenceself-hostedace-step-1-5

DISCOVERED

55d ago

2026-04-03

PUBLISHED

55d ago

2026-04-03

RELEVANCE

8/ 10

AUTHOR

seamonn