← Back to blog

How to Use MiniMax Music: 5 Credits a Generation

The cheapest music on the canvas, and it does lyrics. Where five cents is enough, and where it is not.

MiniMax Music v2 is 5 credits per generation, the cheapest audio on this canvas, and it generates music with lyrics. MiniMax Music 3 is the newer tier at 17.

At five cents a generation, the right way to use it is volume: generate thirty directions, keep one.

TL;DR

The ladder

Canon 8frame prices, a credit is $0.01 at pack rate:

Route Credits Notes
MiniMax Music v2 5 per generation Cheapest, supports lyrics
Lyria 2 13 per 30 seconds Consistent instrumental beds
MiniMax Music 3 17 per generation Newer tier
ElevenLabs Music 81 per minute Premium, for exposed music
ElevenLabs SFX 5 per effect Texture layers

Where five credits is enough

Exploration. Thirty directions for $1.50 to find what a piece should sound like. Nothing else on the canvas lets you be this thorough.

Background beds where nobody is listening closely.

Hooks and stings that will be cut short anyway.

Demos. Sketching an idea before recording it properly, or before briefing a composer.

Lyrics experiments, since it is the option here that produces them.

Where it is not enough

Long-form consistency. A voice or an arrangement that has to hold for three minutes drifts. Lyria 2 at 13 per 30 seconds is more consistent, and the loop approach avoids the problem entirely.

Exposed music. A title sequence, a brand film where the score carries emotion. ElevenLabs Music at 81 a minute.

Anything where the vocal is the product. Generated vocals thin out in the top of the register, blur fast consonants, and lack ad-libs and layering. That is true across the category, not specific to MiniMax.

The loop approach, which applies here too

Generate 20 to 30 seconds and loop it rather than a full track. A 10-minute video needs one 30-second bed, not ten minutes of music. Under narration nobody notices, and generated music often has a dynamic arc that is a problem under speech anyway.

Test the join by playing the end into the beginning ten times, then crossfade half a second in your editor.

Prompting it

Name components rather than genres, and state the function first:

  1. The job. "Background bed under narration, sparse, no lead melody."
  2. Tempo and feel. "90 bpm, laid back, behind the beat."
  3. Named instruments. "Upright bass, brushed drums, Rhodes." This does more than any genre label.
  4. Texture and era. "Warm tape saturation, slight wow and flutter."
  5. What to leave out. "No vocals, no lead line, no big transitions."

For lyrics, write them yourself. Generated lyrics land on cliché reliably, because they are the statistical average of what has been written, and the words are the one part where you have an advantage.

The rule with no exceptions

Never generate a named artist's voice, and never reproduce a recognisable melody. Prompting toward a living performer's vocal identity is a legal exposure and a terms violation on every serious service.

Stylistic reference is different from identity: "1970s soul, warm tape saturation, female lead" is a description. A person's name is not.

Layering, which is the cheap trick

The texture that makes generated music feel real usually is not in the music generation. Layer it:

Music at 5 plus two texture layers at 5 each is 15 credits, fifteen cents, and the layering does more for authenticity than the model choice.

FAQ

How much does MiniMax Music cost? v2 is 5 credits per generation, Music 3 is 17.

Does it do vocals? Yes, it generates music with lyrics. Vocals are the weakest part, as everywhere.

Should I generate a full track? Generate 20 to 30 seconds and loop it unless the music is exposed.

What is the most consistent option? Lyria 2 at 13 per 30 seconds for instrumental beds.


Generate thirty for $1.50, keep one, write the words yourself. The 8frame canvas is free and unlimited, and generation is paid from $19/month.

Related articles

guideHow to Fix Flicker in AI VideoguideMy AI Video Is the Wrong Aspect Ratio. Now What?guideHow to Match AI Clips to Real Footage

Make it
move.

Stay in the loop

Be the first to hear about our launch and get product updates