MiniMax Launches Music 2.6: Chinese-style Tracks Gain "Breathing Rhythm", Game Soundtracks More Dynamic

MiniMax recently released its new music generation model Music 2.6. Ordinary users can generate music 500 times for free daily, while developers get an extra 100 API calls per day.

Previous AI music models often produced Chinese-style tracks without breathing rhythm, feeling like mechanical patches. Music 2.6 understands the logic of traditional opera's opening gongs and drums, allowing string and plucked instruments to enter layer by layer, with melody and vocals building to a climax—giving Chinese-style tracks a real "breathing sense".

In addition, the model optimizes mid-low frequencies for game soundtracks, making drumbeats and bass clearer; supports structured prompts like "from suppression to awakening then to explosion" for music generation; and can create via agent commands such as minimax-music-gen.