MiniMax Music 2.5

• 🎵 Text-to-song model generating full tracks with vocals and instrumentation
• 📏 Accepts up to 3,500-character lyrics with structure tags
• 🎯 14 structural tags: Intro, Verse, Chorus, Bridge, Build-up, Hook
• 🆕 Expanded sound library to 100+ instruments, studio-grade mixing
• 🔧 Instrumental mode plus lyrics optimizer for prompt-based lyrics
• 🎤 Optimized vocal generation with progressive emotional shifts across sections
• 🏢 Built by MiniMax; released 2026, succeeded by Music 2.6

MiniMax

MiniMax is an AI company building generative models across multiple modalities, with a focus that spans both language understanding and audio creation. Their rapid release cadence in early 2026—delivering several new models within just a few months—reflects…

MiniMax Music 2.5 is a text-to-song generation model from MiniMax that turns a style prompt and lyrics into a complete track with synthesized vocals, layered instrumentation, and professional mixing. It supports structure tags, an instrumental mode that produces vocal-free music, and a lyrics optimizer, accepting lyrics up to 3,500 characters.

Compared with its same-family predecessor MiniMax Music 2.0, MiniMax's own announcement frames 2.5 as breaking through two bottlenecks it calls "paragraph-level precision control" and "physical-grade high fidelity." Per that material, the model expands its sound library to more than 100 instruments, adds studio-grade mixing that keeps vocals and accompaniment separated, and opens full-section tag control supporting 14 structural variations including Intro, Bridge, Interlude, Build-up, and Hook. MiniMax also describes systematic optimization of vocal generation, with emotion evolving progressively across sections.

Within the family, 2.5 sits between 2.0 and the newer MiniMax Music 2.6. MiniMax's broader catalog also spans text models and speech synthesis, so Music 2.5 represents the audio-generation branch of that lineup.