Introduction

Artificial Analysis just reset its music leaderboards. On October 7, 2026, it launched AA-Music-Vocal v1.1 and AA-Music-Instrumental v1.1, the two benchmarks behind the Artificial Analysis Music Arena, rebuilt on a brand-new set of 1,000 prompts across 17 genres and scored only by its recruited listening panel.

The headline result: Suno v6 is #1 on both boards, and Suno holds the top two spots on each. This is a benchmark report, not a model drop, so there's nothing new to download, but both leaderboards are public today and you can still cast your own blind votes in the Arena.

Pro Tip

Read the rank ranges, not just the rank numbers. Artificial Analysis publishes a 95% interval for every model (roughly ±14–16 Elo here), so models whose ranges overlap are effectively tied. That's where most of the interesting story in v1.1 lives.

What Changed in v1.1

Per Artificial Analysis's launch thread:

  • New prompt set: 1,000 prompts, 500 per benchmark, balanced across 17 genres from Classical and Jazz & Blues to EDM, Metal, and Reggae.
  • Harder prompts: each one tests song structure, mood, instrumentation, or vocal style, with longer, more complex wording meant to reflect how people actually prompt frontier music models now.
  • Panel votes only: ratings come entirely from blind votes by the recruited evaluator panel. Public Arena votes don't count toward the score.
  • Scale: more than 37,000 blind preference votes at launch (16,000+ on Vocal, 21,000+ on Instrumental), with over 2,000 votes for every ranked model.
  • No lyrics supplied on Vocal: models write their own words, so AA-Music-Vocal measures end-to-end songwriting, not just singing.

Both boards use MiniMax Music 3.0 as the anchor, fixed at 1000 Elo, so every other score reads as "better or worse than MiniMax Music 3.0."

AA-Music-Vocal v1.1: The Top of the Board

Rank (range)ModelElo

1 (1)

Suno v6

1142

2 (2–3)

Suno v6-mini

1116

3 (2–3)

Mureka V9.5

1103

4 (4–6)

Mureka V9

1052

5 (4–6)

Lyria 3.5

1052

6 (4–6)

Lyria 3 Pro

1046

7 (7–11)

Eleven Music v2.5

1012

11 (11)

MiniMax Music 3.0

1000 (anchor)

The standout is Mureka V9.5: it lands 51 Elo above Mureka V9, which Artificial Analysis calls the largest gain between two versions of the same model family on the Vocal board, and its range overlaps Suno v6-mini's for second place.

AA-Music-Instrumental v1.1: A Four-Way Tie Behind Suno

Rank (range)ModelElo

1 (1)

Suno v6

1140

2 (2)

Suno v6-mini

1109

3 (3–6)

Mureka V9

1081

4 (3–6)

Mureka V9.5

1080

5 (3–6)

Lyria 3 Pro

1078

6 (3–6)

Lyria 3.5

1076

8 (7–10)

Stable Audio 3 Large

1017

13 (13)

MiniMax Music 3.0

1000 (anchor)

14 (9–16)

Stable Audio 3 Medium

998

Mureka V9, Mureka V9.5, Lyria 3 Pro, and Lyria 3.5 sit within 5 Elo of each other, a statistical tie for third through sixth. Further down, the instrumental board also gives a useful sense of how far the field has moved: Lyria 2 scores 932 and MusicGen 881, at the bottom of the 19-model list.

The Open-Weights Picture

Artificial Analysis names MiniMax Music 3.0 as the leading open-weights music model, at #11 on Vocal and #13 on Instrumental, with Stable Audio 3 Medium within a few Elo of it on Instrumental (998 vs. 1000). In other words, the best openly available music models currently score roughly 140 Elo behind Suno v6 on both boards, still a clear gap, but they're ranked against the same prompts and panel as the closed leaders.

A Prompt to Try

The v1.1 prompts are long, genre-specific, and describe arrangement and vocal delivery rather than just a vibe. If you want to test your own favorite model the same way, here's a prompt in that style (my own, not one from the benchmark):

Late-night city-pop ballad, warm and nostalgic, around 92 BPM. A soft female
lead sings close to the mic about watching the last train leave without her,
calm on the verses and opening up on the chorus. Clean chorus-pedal electric
guitar, round fretless bass, gentle Rhodes chords, and a tight, dry drum kit
with brushed hi-hats. Verse, pre-chorus, chorus, short saxophone solo, final
chorus with stacked harmonies, then a quiet fade on the Rhodes.

Run it on two or three models without telling yourself which is which, and you'll hear quickly why the panel results are so close in the middle of the pack.

Why It Matters

Music generation has had fewer neutral, public comparisons than image or video, and a lot of model launches arrive with vendor-picked demos. A fresh prompt set, blind panel-only votes, and published confidence ranges make AA-Music v1.1 a cleaner reference point for deciding which model to try for vocals versus instrumentals. It's also a reminder that "newer" isn't always "better on every axis": Mureka V9.5 jumps on vocals but ties V9 on instrumentals, and the closed and open leaders still don't overlap.

Conclusion

Artificial Analysis's AA-Music-Vocal v1.1 and AA-Music-Instrumental v1.1 launched October 7 with 1,000 new prompts, 17 genres, and 37,000+ panel votes. Suno v6 leads both boards, Mureka V9.5 is the big vocal climber, the Mureka/Lyria pack is tied on instrumentals, and MiniMax Music 3.0 is the top open-weights model. The leaderboards are live now, and the Arena is open if you want to add your own blind votes.

Primary:

, AA-Music-Vocal v1.1 leaderboard, AA-Music-Instrumental v1.1 leaderboard, Artificial Analysis Music Arena.

—Anabel ♡