MiniMax Music 3

Produces a full-length track, with vocals or instrumental, from a description of genre, mood, and tempo.

modellane/minimax-music-3 · Text to audio · MiniMax

A model that turns text into a finished music track. The prompt describes genre, mood, tempo, instruments, and vocal colour; the model produces the arrangement, the instrumentation, and the mix together. Lyrics that are supplied get sung; if none are supplied, the model writes its own. With the instrumental option on, the track comes out without vocals. The output is a finished audio file whose format is chosen from the parameter list in the run form — so the track can go straight into an edit.

Vendor

MiniMax

Task

Text to audio

Output

Audio

Price

$0.22 · per run

Input parameters

• Prompt prompt — required, at most 2000 characters Describe what you want produced as concretely as you can.
• Instrumental is_instrumental — optional, default: false When on, the track is produced without vocals and any lyrics are ignored.
• Let the model write the lyrics lyrics_optimizer — optional, advanced, default: true When on, the lyrics are derived from the prompt; the lyrics field must stay empty.
• Lyrics lyrics — optional, advanced, at most 3500 characters Line breaks separate the lines. If you are writing lyrics, turn "Let the model write the lyrics" off.
• File format format — optional, advanced, default: mp3, options: mp3, wav

Recommended use cases

• Music for video and ads — An original bed described to match the clip's length and tempo; no browsing stock libraries.
• Game and app audio — Produce loopable tracks with their own mood for each menu, level, or screen.
• Lyric demo — Hear written lyrics in a vocal arrangement; they are typed into the lyrics field line by line.
• Podcast theme — A short, recognisable theme for the intro and outro; with the instrumental option it can also run under speech.

Where it is strong

• Produces arrangement, instrumentation, and mix in a single step
• Sings the lyrics it is given, or writes its own when none are
• The instrumental option yields a track that can run under speech
• Delivers a single mixed audio file that can go straight into an edit

Limits

• Does not imitate a particular artist's voice or an existing song
• Gives no score, chord chart, or separate instrument tracks; the output is one mixed piece
• You cannot specify which second a line is sung at
• The same prompt produces a different arrangement on every run

Example prompt

Yavaş tempolu lo-fi hip hop; sıcak Rhodes akorları, boğuk davul, hafif vinil çıtırtısı, gece boyunca çalışmaya eşlik eden sakin bir ruh hâli

Example request

POST /v1/queue/modellane/minimax-music-3

Playground

The identifier, price, parameters, and example requests on this page can be read without signing in. Running the model from the panel sends a real generation request and charges the project balance; the playground therefore requires an account.