The music generation AI ACE-Step 1.5 XL was released as an open model on April 2, 2026, marking a significant upgrade from its predecessor, ACE-Step 1.5, which debuted in February 2026. Designed to generate songs featuring Japanese vocals, ACE-Step 1.5 XL can be operated on local PCs, expanding accessibility for creators and developers alike.
Developed with a robust architecture, ACE-Step 1.5 XL utilizes a 4 billion parameter DiT model, enhancing audio quality and musicality while maintaining rapid generation speeds. The model’s ability to follow prompts closely allows users to generate music by providing simple instructions in natural language. A demonstration page featuring songs created with Japanese vocals is accessible, showcasing the technology’s capabilities.
According to a tweet from ACE Music, the AI has been scaled to improve its performance, stating, “We scaled the DiT decoder to 4B. And it shows better audio quality, better prompt following, and better musicality.” The tweet highlighted that while the model’s speed remains impressive at just eight steps with turbo distillation, the core generation API and training code have not changed, ensuring familiarity for existing users.
ACE-Step 1.5 XL is available for download in three distinct versions: the base model acestep-v15-xl-base, the fine-tuned model acestep-v15-xl-sft, and the distillation model acestep-v15-xl-turbo. Each version has specific strengths; the base model is noted for high diversity, while the fine-tuned version provides improved quality with a slight trade-off in diversity. The turbo model prioritizes speed and high-quality audio but offers limited fine-tuning capabilities.
The minimum VRAM requirement for running ACE-Step 1.5 XL is 12GB, with a recommendation of 20GB or more for optimal performance. The differences between the models are clearly delineated; for instance, the base model supports a total of 50 steps in generation, yielding medium quality, while the fine-tuned variant also allows 50 steps but achieves high-quality results. The turbo model, in contrast, operates with just eight steps, providing very high quality but at the expense of diversity and fine-tuning capability.
As part of its commitment to user support, ACE-Step 1.5 XL retains workflows familiar to users of its predecessor. A comprehensive Japanese tutorial titled ‘ACE-Step 1.5 Ultimate Guide (Must Read)’ is also available, providing essential insights into maximizing the model’s capabilities.
The introduction of ACE-Step 1.5 XL reflects ongoing advancements in AI-driven music generation and highlights the increasing demand for tools that facilitate creative expression. As developers continue to refine and enhance AI technologies, the potential applications in various creative industries appear boundless, opening new avenues for artists and musicians in the digital landscape.
For more information and to access the models, visit the following links: ACE-Step/acestep-v15-xl-base, ACE-Step/acestep-v15-xl-sft, and ACE-Step/acestep-v15-xl-turbo on Hugging Face.
See also
Germany”s National Team Prepares for World Cup Qualifiers with Disco Atmosphere
95% of AI Projects Fail in Companies According to MIT
AI in Food & Beverages Market to Surge from $11.08B to $263.80B by 2032
Satya Nadella Supports OpenAI’s $100B Revenue Goal, Highlights AI Funding Needs
Wall Street Recovers from Early Loss as Nvidia Surges 1.8% Amid Market Volatility
