ACE Music v1, the song-generation model from China's ACE Studio, has taken first place in Artificial Analysis' AI vocal music ranking, with 1,160 Elo points — 16 ahead of Suno v6, the category's leader until now. The result, recorded on November 11, 2026 in the AA-Music-Vocal v1.1 ranking, is the first time any model has sat above Suno v6 since that version launched on September 9, 2026.
The lead, however, is narrow. The two models' 95% confidence intervals overlap — 1,139 to 1,181 for ACE Music v1 and 1,126 to 1,162 for Suno v6 — and Artificial Analysis itself gives ACE a rank range of 1st to 2nd. In practice it is a statistical tie at the top: 16 Elo points amount to roughly a 52% to 48% edge in a blind head-to-head. ACE has 2,119 comparisons on the board, against 3,063 for Suno v6. In third place sits Suno v6-mini, with 1,132 points.
How the ranking works
AA-Music-Vocal v1.1 evaluates AI-generated songs with vocals. Evaluators listen to pairs of tracks made from the same prompt, without knowing which model made each one, and pick their favorite. Those preferences become an Elo-style score: the higher it is, the more often the model was preferred.
There is an important limit to ACE's achievement: on the separate instrumental music board, Suno v6 still leads with 1,148 points, and ACE Music v1 does not appear on the list at all. For now, the lead applies only to songs with vocals.
What ACE Music v1 is
ACE Studio was founded in 2018 and spent years building AI singing voices from MIDI notes and lyrics. ACE Music v1 is its first full-song model: it generates complete songs from text descriptions or lyrics supplied by the user.
Its Idea feature develops an initial concept into lyrics and a structured composition, identifying the song's sections and how they should be arranged, so the user can refine the material instead of generating everything again. A preview of version 1.1 goes further and creates a musical score before generating audio: users can edit melodies and chords, regenerate performances, export MIDI files and separate vocals from instruments.
Hands-on test: better melodies, weaker niches
In hands-on tests, Decrypt found more melodic and coherent songs in mainstream styles, with more natural voices and cleaner endings than Suno — which sometimes introduces strange sounds, repetitions or abrupt changes near the end. In less conventional genres, though, Suno had the edge: in a melodic death metal test, ACE kept a clean, pleasant female voice, missing the aggressive delivery expected from the style. The tests themselves were subjective, not a controlled benchmark — and they show that topping the leaderboard does not make one model better for every musician.
Access and open-source plans
The flagship model is proprietary and has no public API: it can only be used in the ACE Music web app, with a limited free plan and paid subscriptions starting at $7.90 per month on annual billing. Downloads and commercial use of generated songs require a paid plan.
At the same time, the company keeps an open-source strategy: ACE-Step 1.5, a separate and weaker family of models, already has public code and weights and can run on users' own hardware. And the developers state on the official site that they plan to open-source ACE Music v1 Flash later in 2026 — a version designed to balance quality, generation speed and hardware requirements.
Image credit
Image credit: screenshot of the Artificial Analysis AA-Music-Vocal v1.1 leaderboard table, published by Decrypt.
Sources
- Decrypt — This New AI Music Model Is Already Better Than Suno
- Artificial Analysis — AA-Music-Vocal v1.1 Leaderboard
- ACE Music — official site
- OrcaRouter — ACE Music v1 Tops AA-Music-Vocal
Transparency: this article was produced with the help of artificial intelligence, based on public posts on X and sources on the internet, and reviewed before publication. Please check the original sources to confirm the information.
By Muse