MiniMax released Music 2.6 on April 10, adding a Cover feature that extracts a song's melodic skeleton and lets creators rebuild everything around it. Style, arrangement, lyrics, and genre are all interchangeable while the original melody stays intact.

What Happened

Music 2.6 ships with three headline changes. The Cover feature takes an existing track, isolates its melodic core, and regenerates the surrounding elements. A folk song can become heavy metal. A classical arrangement can shift to cyberpunk electronic. Lyrics can be replaced entirely while the melody holds.

Alongside Cover, MiniMax cut first-packet latency to under 20 seconds and overhauled the model's low-frequency acoustics. Sub-bass and drum presence are noticeably tighter, and the model now handles layered instrument entry with performance nuances like vibrato and breath pauses across 100+ instruments.

Why It Matters

Cover mode moves AI music generation from blank-canvas creation into remix and adaptation territory. For creators working with existing compositions, this opens up genre adaptation, localization through lyric replacement, and rapid style exploration without losing the melodic identity of a track. The sub-20-second generation time makes it practical for iterative creative workflows rather than batch-and-wait production.

Key Details

  • Cover mode preserves melody while transforming style, arrangement, genre, and lyrics
  • First-output latency under 20 seconds
  • Enhanced BPM, key, song structure, and emotional arc control in prompts
  • 500 free creations per account daily during the 14-day global beta
  • Full API access for both Music 2.6 and Cover features, with 100 bonus API calls per day for token plan users

MiniMax also open-sourced three agent-native music tools: minimax-music-gen for automated generation, minimax-music-playlist for taste-based curation, and buddy-sings for character-persona song improvisation.

What to Do Next

Try Cover mode at minimaxi.com/audio/music during the free beta window. The API documentation covers both generation and Cover endpoints for developers building music into workflows. MiniMax's CLI tool also supports music generation for terminal-based pipelines. For context on the broader AI music landscape, see our coverage of ElevenLabs ElevenMusic.