MiniMax Music 2.6 — Full Songs, With or Without Your Lyrics
Singing, Backing Music, and Arrangements from a Style Description
MiniMax Music 2.6 creates complete tracks with singing, backing music, and detailed arrangements from lyrics and a style description. What's new in this generation is flexibility in how the words get written: supply your own lyrics (up to 3,500 characters, with structure tags like [Intro], [Verse], and [Chorus]), leave the lyrics field empty and the built-in lyrics optimizer writes them from your prompt, or flip on instrumental mode for a track with no vocals at all. The style prompt accepts up to 2,000 characters — enough room for genre, era, tempo, vocal character, and scene. Output is MP3 at 44.1kHz/256kbps by default, and commercial use is supported.
Compared with MiniMax Music V2 on this site, the 2.6 endpoint changes the input contract in three documented ways. First, lyrics are optional: when the lyrics field is empty and instrumental mode is off, the lyrics_optimizer writes lyrics from your style prompt — useful when you have a vibe but no words. Second, there is a true instrumental switch (is_instrumental) for generating vocal-free tracks, which V2 does not expose. Third, the style prompt limit grows from 300 to 2,000 characters, so you can describe arrangement details, vocal character, and scenario at length — the official example reads 'City Pop, 80s retro, groovy synth bass, warm female vocal, 104 BPM, nostalgic urban night'.
When you do bring your own lyrics, the same structure-tag convention applies: [Intro], [Verse], [Chorus] and similar tags mark sections, one line per lyric, up to 3,500 characters. Audio settings are documented at sample rates from 16kHz to 44.1kHz and bitrates up to 256kbps, in MP3, WAV, or PCM — we use the defaults (MP3, 44.1kHz, 256kbps).
Generation runs asynchronously: progress shows on the page, results land in your dashboard gallery, and the flat per-track credit cost is displayed on the Generate button before you submit.
How to Generate a Song with MiniMax Music 2.6
Describe the Style
Write a style prompt up to 2,000 characters — genre, era, tempo, vocal character, mood. Example from the official docs: 'City Pop, 80s retro, groovy synth bass, warm female vocal, 104 BPM'.
Choose How the Words Happen
Paste your own lyrics with [Verse]/[Chorus] tags, leave the field empty to let the model write them, or switch on instrumental mode for no vocals.
Generate & Download
Click Generate and preview the finished MP3 in the player. Every generation is also saved to your dashboard gallery.
MiniMax Music 2.6 Technical Specifications
| Provider | MiniMax (fal.ai partner) |
| Platform | fal.ai |
| Style Prompt | 10–2,000 characters |
| Lyrics | Optional, up to 3,500 characters, structure tags supported |
| Auto Lyrics | lyrics_optimizer writes lyrics from the prompt when the field is empty |
| Instrumental Mode | Yes (is_instrumental) |
| Audio Output | MP3, 44.1kHz, 256kbps (default); WAV/PCM available in the API |
| Commercial Use | Supported |
| Processing | Asynchronous (queue-based) |
Why Choose MiniMax Music 2.6
Lyrics Are Optional
Bring your own words, or leave the field empty and the built-in lyrics optimizer writes them from your style prompt — the fastest path from idea to finished song.
True Instrumental Mode
The documented is_instrumental switch produces vocal-free tracks — something MiniMax Music V2 doesn't offer.
Room to Direct
A 2,000-character style prompt leaves space for genre, arrangement, vocal character, tempo, and scene — far beyond V2's 300-character limit.
MiniMax Music 2.6 vs MiniMax Music V2 vs ElevenLabs Music
| Feature | MiniMax Music 2.6 | MiniMax Music V2 | ElevenLabs Music |
|---|---|---|---|
| Style Prompt | 10–2,000 characters | 10–300 characters | Free-form text |
| Lyrics | Optional, up to 3,500 chars | Required, 10–3,000 chars | Optional, via composition plan |
| Auto-Written Lyrics | Yes — lyrics optimizer | No | No |
| Instrumental Mode | Yes (is_instrumental) | No | Yes — guaranteed toggle |
| Default Output | MP3, 44.1kHz/256kbps | MP3, 44.1kHz/256kbps | MP3, 44.1kHz/128kbps |
| Length Control | Model decides | Model decides | Exact, 3 seconds–10 minutes, or auto |
What Can You Create with MiniMax Music 2.6
Songs Without Songwriting
Describe the mood and let the lyrics optimizer write the words — a complete sung track from a style description alone.
Original Songs from Your Lyrics
Bring finished lyric sheets with structure tags and hear them performed with backing music and detailed arrangements.
Instrumental Versions
Flip on instrumental mode to get a vocal-free rendition — useful for backing tracks and music under narration.
Richly Directed Productions
Use the long prompt to specify era, instrumentation, BPM, and vocal character in one pass, like the official City Pop example.
Content Theme Songs
Create signature songs for channels and series, with or without your own written hooks.
Songwriting Drafts
Test how your half-written lyrics sound when performed, then iterate on the words based on what you hear.
Related AI Models
MiniMax Music V2
Full songs with vocals from your own lyrics — verse, chorus & bridge control
ElevenLabs Music
Studio-grade AI music from a text prompt — vocals or instrumental, up to 10 minutes
Sonilo v1.1
Production-ready music from one prompt — exact duration control, up to 10 minutes
CassetteAI
Lightning-fast music generation — a full 3-minute track in seconds
Frequently Asked Questions About MiniMax Music 2.6
Do I have to write lyrics?
No. Lyrics are optional in 2.6: if you leave the field empty (and instrumental mode is off), the documented lyrics_optimizer writes lyrics based on your style prompt. If you do provide lyrics, they can run up to 3,500 characters with structure tags like [Intro], [Verse], and [Chorus].
Can it generate music without vocals?
Yes. MiniMax Music 2.6 documents an is_instrumental option that generates vocal-free music. On this page it's a single toggle — when enabled, the lyrics field is hidden and no vocals are produced.
How is 2.6 different from MiniMax Music V2?
Three documented differences: the style prompt grows from 300 to 2,000 characters; lyrics become optional with an automatic lyrics optimizer; and there's a dedicated instrumental mode. V2 remains available on LoveGen if your workflow is built around required lyrics.
What audio format do I get?
MP3 at 44.1kHz and 256kbps by default. The API also documents WAV and PCM output and sample rates from 16kHz to 44.1kHz; this page uses the defaults. Download from the player or your dashboard gallery within 7 days.
Can I use the songs commercially?
Yes. MiniMax Music 2.6 is marked for commercial use on fal.ai, so generated tracks can be used in monetized content, ads, client projects, games, and releases.
How much does a song cost?
Each generation costs a flat number of credits regardless of song length, reflecting the provider's flat per-track pricing. The exact amount is shown on the Generate button before you submit, and failed generations are automatically refunded.