Hands-on review

Suno v6 review: great sound, selective hearing

Oct 6, 2026 · 8 prompts, 16 takes, scored blind by one listener.

The review in a song: "Selective Hearing," made with Suno v6 from our lyrics.

Verdict

Suno v6 launched Sep 9, 2026, and retired every earlier Suno model (Suno v6 FAQ, release notes). Some users haven't been kind. MusicRadar reported a backlash from subscribers calling it "muffled, generic" (MusicRadar, Sep 17, 2026).

So we ran a blind test. Nathan Crowe, our editor, has made more than 1,000 AI tracks, which he hosts on Audius. He scored all 16 takes without seeing the prompts or settings. The average was 4.56 out of 5, with ten 5s and nothing below a 3. Nothing in his notes mentions a muffled mix.

The sound is good. The mashups are better than they have any right to be. Tagged structure switches landed. What v6 still doesn't do reliably is follow specific production instructions: a key change, an instrumental's length, a build that saves the full band for the end. It also still reaches for the same sound when a prompt leans on a loaded keyword. In Nathan's experience, neither problem is new. He's hit both since v2. v6 hasn't fixed them.

Verdict: v6 sounds good enough that the "muffled, generic" complaints didn't come up in our blind test, and it handles wild genre mashups and tagged structure switches. It still ignores specific production instructions and repeats itself on loaded keywords, as Suno has since v2. Use it for the sound, and plan on rerolls for the details.

• Who it's for: songwriters and hobbyists who want strong-sounding demos, odd genre blends, and songs with clear sections, in two takes for 10 credits.
• Who should skip it (for now): anyone who needs a song to hit an exact length, key change, or arrangement on the first try, or who isn't comfortable with the legal questions below.

Key facts

Hands-on test

What we measured: whether each song did what the prompt asked, and how it sounded. Listening is one person's ear: Nathan Crowe scored every take blind from the links alone, then answered follow-up questions about the instructions after we unblinded. One generation per prompt, so treat this as a sample, not a benchmark.

Setup: suno.com web app, Oct 3, 2026. Model v6 standard, Max Mode off. All 8 generations ran between 10:24 and 10:36 AM CT. Credits went from 4,755 to 4,675: 80 credits for 8 generations, 10 each, 2 takes each. Lyrics were original, written for this test, with no real artist names. Takes are listed in the order Suno returned them.

Scores and instructions

Tally: 73 points across 16 takes, an average of 4.56. Ten 5s, five 4s, one 3. Take B beat Take A in five of eight pairs and tied in the other three. Take A never won.

1. Paper Plane Season (easy)

Style: upbeat indie pop, female vocal, handclaps, bright clean guitar, 118 BPM

The warm-up. Nathan heard Take A as a catchy, kind-of-indie pop singer (4). Take B started out the same but had a better chorus, with "a little bit of a... mournful? wishful? atmosphere" (5). Both takes are short, at 1:23 and 1:24. We didn't ask for a length here, so we won't hold that against it.

2. Rain Car (easy, Simple mode)

Description: a garage rock song about an old car that only starts when it rains

We gave Suno one sentence. It wrote the lyrics and titled the takes "When the Rain Comes Down" and "It Only Starts When It Rains." Both stayed on the car-that-only-starts-in-the-rain story, which is more than some human songwriters manage. Take A was groove-driven rock with a decent male vocal and a catchy chorus (3, the lowest score of the test). Take B was rawer and more punk, shifting into a big alternative chorus (4).

3. Thermos (medium)

Style: acoustic country ballad, male baritone, fingerpicked guitar, pedal steel, 72 BPM, sparse at first, full band joins on the last chorus

Blind, Nathan loved these. Take A had "good twang and drawl" and "feels very 90s era country, not like the current style" (4). Take B had the same feel and a more powerful chorus (5).

Then we asked about the one instruction. Take A's last chorus sounded about the same as its first, if anything softer. Take B brought the full band in on the first chorus, and its last chorus was almost vocals only, with the drums dropping out. That's the instruction in reverse. It sounded great. It just wasn't what we asked for.

To be fair to v6, our first lyrics used a bare [Chorus] tag in the middle, so we reran it with every chorus written out and the same settings. Every chorus got sung this time. The build still didn't happen. On Take A, the second and third choruses came out almost equally full. On Take B, the second chorus was the biggest, and on the last one the drums dropped out, leaving mostly vocals. That's the same reversal, twice.

4. Municipal (hard)

Style: boom bap hip hop, male rapper, fast double-time flow, dusty drums, vinyl crackle, 92 BPM

Sixteen bars of dense, multisyllable lines about city hall. Nathan's verdict: it "nails every word both times." Take A was "very 90s hip hop... boom bap... dude can flow" (5). Take B worked the old-record sound into the chorus, "jumping from live vocals to samples from a record" (5). Both takes are short (1:36 and 1:26), but nothing in the brief asked for length, and the lyrics were delivered.

5. Holler Code (hard)

Style: bluegrass banjo and fiddle over drum and bass breakbeats, 174 BPM, female vocal with Appalachian twang

The standout. Mashup prompts usually get you one genre with a hat from the other. Not here. Take A builds from bluegrass until the chorus hits, and then, in Nathan's words, "bluegrass and 2000s electronic pop music had a baby," with a southern drawl over digital beats (5). Take B is "totally drum and bass... bluegrass... wild," and "bizarrely catchy" (5). His summary, which we can't improve on: "my compliments to the chef."

6. Music Box Riot (hard)

Style: starts as a slow 3/4 waltz with music box and upright bass, soft female vocal; switches to fast punk rock at 180 BPM with shouted gang vocals; returns to the slow waltz at the end

Two hard switches, tagged in the lyrics, and both takes made both of them: waltz, punk, and back to the waltz at the end. Take A opened like a waltz, then went to a screaming and singing female chorus that Nathan called "weird and neat" (4, "for uniqueness"). Take B went circus-like, a bit metal, almost jazzy, then "punk as f***" (5). In his words: "a music box, waltzing and punk rock should not work so well together."

So v6 can follow big structural moves tagged in the lyrics. Hold that thought.

7. Tin Roof (hard, instrumental)

Style: instrumental, cinematic ambient score, rain on a tin roof, low cello drones, distant felt piano, slow build over four minutes, no vocals

The current interface had no separate Instrumental toggle. We left Lyrics empty, as the field told us to, and Suno labeled both takes instrumental.

Take A had "great ambience" and would work for a cinematic sequence (4). Take B was "very beautiful," and Nathan may use it as B-roll music in a video series he's editing (5).

Two catches. First, after we unblinded, Nathan said there was "maybe some humming," very soft and buried in heavy reverb, "like a church group singing maybe, at Christmas," before the instruments took over. Our rule for this test was that any singing or humming is a fail. He heard it in both takes, so Tin Roof fails the instrumental test, softly but clearly. Second, we asked for a build over four minutes. The takes ran 3:49 and 3:17.

The bigger complaint was about Take A: it sounded too much like a track Nathan made himself in Suno weeks ago (listen here). Same kind of keywords, same kind of sound. More on that below.

8. Last Ferry Out (hardest)

Style: power ballad duet, deep male vocal and high female vocal trading verses, piano and strings, full band from the first chorus, over 4 minutes, key change up a whole step on the final chorus

Getting a male and a female voice to take turns has been a sore spot. One Suno user told us he regenerated more than 30 times trying to get the voices to sound different. v6 did it on the first try, twice. Take A was "SO 80s," with a powerful male vocal answered by a female one (5). On Take B, "the duet is even better... getting chills... banger" (5).

Then the fine print. Neither take had a key change Nathan could hear on the final chorus, which was "just more powerful in both instances." Neither ran over 4 minutes (3:22 and 3:24), but that one's on us. We didn't write four minutes of lyrics, and Suno sings what's on the page.

So we reran it with longer lyrics, every chorus written out, and the same settings. Both takes cleared four minutes (4:08 and 4:27), and the duet got better. Take B leaned more orchestral, the voices overlapped in interesting ways, and the bridge was better arranged than in the first round. The key change still didn't happen. Both takes read the [Key Change] tag as a cue to drop to soft vocals and piano, then came back big in the same key. That makes it 0 for 4.

The old problem v6 didn't fix

The fails in this test line up. v6 followed the big moves: a switch from waltz to punk and back, and whose turn it was to sing. It ignored the finer production notes: a key change, an instrumental's length, when the band comes in. Writing them in twice didn't help, and neither did a rerun with every chorus written out. The key change was in both the style box and a [Key Change] lyric tag. The full band was in both the style box and a [Final Chorus - full band] tag. v6 treated both as suggestions, and in one take it did the reverse.

The second problem is sameness. Some keywords pull Suno toward the same result every time. Nathan's examples are from his own work, not from this test. With "noir" and "jazz" in a prompt, Suno gave him the same melody 17 generations straight, all on v4.5 in one evening in May 2025. Hear three of them side by side: one, two, three. "Grunge" or "Seattle grunge" almost always defaults to the same kind of guitar riff. Tin Roof's Take A falling close to one of his own ambient tracks is the same problem at small scale. v6 did it in our test too. All three Thermos takes on this page, two from the first round and one from the rerun, open with the same guitar intro, note for note as far as Nathan can hear. That's three takes across two generations, all from the same style prompt, and the rerun even had different lyrics. Scroll up and play the first few seconds of each.

Neither of these started with v6. Nathan has hit both going back to v2. What's notable is that v6, which retired every earlier model, still has them.

How to get around it: describe the feeling, not the genre label. Instead of "noir," Nathan writes mood words like "forlorn, haunting, melancholy." The label pulls Suno back to its habit, while the mood words leave it room to find something new. Tags like [Tempo Change - Fast Punk] worked for big section switches in our test. For key changes, length, and dynamics, budget for rerolls. One idea we didn't test here: our prompt asked for a key change "up a whole step" without naming a key. In his own songs, Nathan has had better luck naming the key or scale for each section, like a verse in E minor tetratonic shifting to an A hexatonic chorus. Polyrhythms, he says, are harder to get. One more from Nathan: write out every repeated chorus in full. A bare [Chorus] tag tells Suno which melody to use, not what to sing, and an empty one tends to get skipped. Some of our own prompts used bare tags, which may explain a few of the shorter takes. When we reran two tests with full lyrics, the songs got longer, but the key change and the build still didn't happen.

Pricing (checked Oct 2, 2026)

What v6 costs: 10 credits per standard generation, which gets you two takes. That's the same as earlier models (Suno v6 FAQ). Our 16 takes cost 80 credits. Max Mode "costs more credits," and Suno hasn't published the figure. We didn't test it.

Download counts reset on your billing date, and unused downloads don't roll over. A song counts once, whatever the format, and re-downloads and stems don't count again (Suno downloads FAQ). Streaming, sharing, and embedding are unlimited on every plan. "The limits only apply to downloading files off the platform" (Suno downloads FAQ).

Stems: Auto Split (up to 12 stems) costs 50 credits; Split from Mix costs 10 credits per stem (Suno help). We didn't test stems.

Rights: download it or you can't sell it

Suno's terms changed on Sep 3, 2026 (Suno). Pro and Premier subscribers are assigned Suno's rights in their output, but "Suno makes no representation or warranty to you that any copyright will vest in any Output" (Suno terms).

The practical change is downloading. You can commercially use a song only if you've downloaded it within your plan's allowance: "You may not commercially exploit Output that has not been downloaded by you through an approved channel" (Suno terms). Once a song is downloaded, those rights are permanent. They aren't affected if you later run out of downloads, downgrade, or cancel.

So on Pro, a song you want to sell, license, or use in a monetized video uses one of your 20 downloads a month, and you have to download it while you're subscribed. Songs made with Suno's "allow others to remix" feature are personal, non-commercial use only, whatever your plan (Suno terms). This is our reading of the terms, not legal advice.

The lawsuit

v6 launched with Warner Music Group and BMG as partners (release notes). Warner settled with Suno in November 2025 (WMG). The other two majors didn't. On Sep 18, 2026, Sony Music and UMG sued Suno in federal court in Massachusetts. They claim v6 still infringes because it was trained partly on output from the earlier models, "fruit of the same poisoned tree." The suit names 60,202 recordings (Variety, MBW). In a statement responding to the suit, Suno said v6 "was trained on content licensed from our partners, interactions including creations and preference signals from our community, and the accumulated learnings from our team." That appears to be the first time Suno has publicly said songs its users made went into v6's training (MBW). If you make songs on Suno, its terms give it a "perpetual, irrevocable" license to what you create, including for improving "the artificial intelligence and machine learning models." Suno also faces a $1 billion suit from Round Hill and a proposed class action.

None of that changes how v6 sounds. It's worth knowing before you build a release plan on it.

All 16 takes are in one public playlist: Suno v6 Review: Great Sound, Selective Hearing.

Pros and cons

Pros

• Strong sound across genres. 4.56/5 average over 16 blind takes, with nothing below a 3 (our test).
• Genre mashups that actually blend: bluegrass over drum and bass worked in both takes (our test).
• Followed two hard tempo and genre switches tagged in the lyrics (our test).
• Clear diction on dense, fast rap (our test).
• Two clearly different voices in a duet, first try (our test).
• Same price as before: 10 credits for two takes (Suno v6 FAQ).

Cons

• Ignores specific production instructions in the style box: no key change, an instrumental under the requested length, and dynamics backwards in one take (our test). Nathan has seen this since v2.
• Loaded keywords pull it toward the same melody or riff (see above).
• Faint humming in an instrumental, in both takes (our test).
• Commercial use requires a download, capped at 20 a month on Pro (Suno terms, downloads FAQ).
• Being sued by two of the three major labels over this model (Variety).

FAQ

Is Suno v6 muffled? Some users told MusicRadar it is (MusicRadar). In our blind test, Nathan didn't call any of the 16 takes muffled and averaged 4.56/5. That's one listener and 16 takes.

Can I still use v5 or v5.5? No. Every model before v6 is retired, and extending, covering, or remixing an old song now runs on v6 (Suno v6 FAQ).

Does v6 do key changes? Not in our test. We asked in the style box and with a [Key Change] lyric tag, across four takes in two rounds, and Nathan didn't hear a key change in any of them. In the rerun, v6 treated the tag as a cue to go quiet instead.

Can I use v6 songs commercially? On Pro or Premier, yes, but only songs you've downloaded within your plan's allowance, since Sep 3, 2026 (Suno terms).

Sources

All checked Oct 2–3, 2026 (CT).

1. Suno, v6 FAQ (edited Sep 9, 2026)
2. Suno release notes
3. Suno pricing
4. Suno Terms of Service (effective Sep 3, 2026)
5. Suno, "Updates to our terms" (Aug 10, 2026)
6. Suno help, downloads FAQ
7. Suno help, stems
8. Music Business Worldwide, Suno limits subscriber downloads per month
9. Warner Music Group, Warner Music Group and Suno partnership (Nov 2025)
10. Variety, Sony Music and Universal Music sue Suno over label-backed model (Sep 2026)
11. Music Business Worldwide, Suno fights to keep Warner settlement terms away from UMG and Sony
12. MusicRadar, Suno users slam its v6 update (Sep 17, 2026)
13. Song Tested hands-on test of Suno v6: suno.com web app, Oct 3, 2026, v6 standard, Max Mode off; blind scores by Nathan Crowe