Suno v6 Prompt Guide: briefs the new models actually understand

The three v6 models — and when to use each

Suno ships v6 as three variants. Your choice changes how much control you get and how predictable the result is.

• v6 — the flagship. Reliable, precise, consistently polished across genres. Use it when you know the sound you want.
• v6-wild — less predictable and more varied. Excellent for idea generation and happy accidents; refine the keeper back in regular v6.
• v6-mini — the free, fast model. Prioritise your three strongest decisions and let the model fill the rest.

In Suno Prompt Builder you pick the target model in the Prompt Output header, and the brief is written for that model automatically.

Natural-language briefs beat tag stacks

The single biggest v6 shift: the model connects plain musical language to vocals, instrumentation, structure and feel. Stacking loose genre tags (“cinematic, epic, orchestral, trailer, hybrid”) works far worse than describing how the track should sound.

The 9-field brief formula

Every field is optional — the goal is musical decisions without contradictions. A reliable order:

1. Genre and context — primary genre, sub-genre, cultural tradition (Turkish makam, world regions…).
2. Mood — one or two words that anchor the feeling.
3. BPM and groove — v6 holds a locked groove reliably; decide it early.
4. Vocal character — type, range, style, timbre.
5. Core instruments — give each one a role, not just a name.
6. Song structure — how sections should build.
7. Arrangement movement — the verse-to-chorus emotional arc.
8. Production and mix character — warm analog, wide cinematic, raw live room…
9. Focused exclusions — one specific “no”, not a ban list.

Structure tags still matter — keep them clean

Section tags like [Verse], [Chorus], [Bridge] remain the reliable way to shape lyrics. v6 adds plain-language section editing: you can re-edit one section (“turn the chorus into a gospel choir”) without touching the rest — which only works when your structure tags are clean and unambiguous.

Mashups and sampling: assign roles

v6 can combine sources in one request and isolate a riff by timestamp. The rule that separates usable results from mush: give every source an explicit musical job. “Vocals from clip A, drums from clip B, new oud melody over both” beats “blend these two” every time.

Multimodal starts

v6 accepts images, video and voice memos as prompts. Translate what you see into what you hear: a rain-soaked neon street becomes “wet reverb-drenched synths, distant sirens, slow half-time groove” — the model reads the feeling, not the scene.

Quick checklist

• Pick the model first: v6 for control, v6-wild for experiments, v6-mini for speed.
• Describe sound, don’t stack tags.
• Lock tempo and vocal early.
• One focused exclusion (“no excessive vocal runs”) beats a blacklist.
• Keep structure tags clean for section editing and lyric swaps.

Build all of this into a structured brief in the Suno Prompt Builder — the v6 model selector writes the brief for you. Related: v5.5 prompt guide · lyrics guide.