TL;DR - Suno v5.5 (March 2026) is the most expressive model yet, and three personalisation features finally make the platform usable as a real workflow: Voices (clone your own verified singing voice), Custom Models (fine-tune v5.5 on your own catalogue), and My Taste (lightweight preference learning for everyone). The Warner Music deal is now visible in the product - older models are being deprecated, free accounts have lost commercial download rights, and the ownership language has softened from “you own this” to “you have commercial rights.” Best used for demos, stem libraries, and personal sound signatures; still risky for releases that need clean copyright provenance.

Suno has had a busy six months. The v5.5 model shipped on 26 March 2026, the Warner Music Group partnership announced late last year is now reshaping the product, and Suno Studio has matured from a launch curiosity into a workflow people actually use. Here is what the platform looks like in May 2026, what is genuinely useful, and what to watch.

The v5.5 model

v5.5 is the headline release of the year so far. Suno is calling it their most expressive model yet, and the practical difference is that vocal performances no longer fall apart on the awkward edge cases - long held notes, intimate dynamics, breath between phrases. The official help article lays out the changes in full, but three features are doing most of the work.

Voices

Voices is the feature the community has been asking for since v3. You record or upload a clip of yourself singing, Suno verifies it by asking you to sing a randomised phrase, and from then on you can generate songs in your own vocal identity. The verification step exists to stop people cloning a singer they do not have rights to, and Voices are private to your account by default.

This matters because the previous “persona” system - covered in the Voice Personas post - kept consistency across generations but used Suno’s own synthesised voices. Voices closes that gap: the singer can be you. It is gated to the Pro and Premier tiers.

Custom Models

Custom Models let you fine-tune v5.5 on your own catalogue. Upload a handful of original tracks and you get back a personalised version of the model that has internalised your sense of arrangement, harmony, and texture. Pro and Premier accounts can keep up to three custom models active.

The interesting use case is not “make my next song sound like my last one” - it is using a custom model as a brief. If you have spent years developing a sound, you no longer have to write a 200-word prompt to coax Suno towards it.

My Taste

My Taste is the lightweight version of the same idea, available to everyone including free users. It quietly learns the genres, moods, and arrangement choices you keep coming back to and weights generations accordingly. It is the kind of feature you only notice after a few weeks, when prompts start landing closer to the first attempt.

The Warner Music deal is now visible in the product

The Warner Music Group settlement and partnership was announced in November 2025 and has been working its way through Suno’s policies and product over the spring. A few changes are now live or imminent.

  • Older models are being phased out. When the next licensed models ship later in 2026, the current pre-deal models are scheduled for deprecation. If you have legacy projects depending on a specific model version, archive your stems now.
  • Free accounts can no longer download for commercial use. Tracks generated on a free account remain non-commercial, and - critically - upgrading later does not retroactively unlock commercial rights for those tracks. If you want to release a song, generate it on a paid plan from the start.
  • Ownership language has shifted. Suno’s terms used to say subscribers owned the songs they generated. The current wording is more careful: paid users get commercial use rights, but Suno does not characterise them as the “owner” of the output. This is a meaningful distinction for anyone planning to register copyrights or sync the music to film.
  • Songkick is now part of Suno. Suno acquired the Songkick live-music discovery platform from Warner as part of the deal. The product still runs standalone, but it gives Suno a route from “AI-generated track” to “live performance discovery” that no other generative music company has.

The trade-off behind all of this is that the next-generation Suno models will be trained on licensed catalogue with artist consent, and artists named or referenced in prompts can opt in or out at the catalogue level. That is a real change from the v3 and v4 era, where the training data question was largely unanswered.

Suno Studio in May 2026

Suno Studio - the in-browser generative DAW launched in September 2025 - has been the quiet winner of the Warner deal. The new download policy keeps unlimited stem exports inside Studio while tightening them on the standalone song generator, which has pushed more serious users into the Studio workflow. In practice that means: generate or import stems, arrange on a timeline, export to Ableton Live or Logic Pro for mixing.

The combination that finally feels coherent is Voices + Custom Model + Studio: a consistent singer, a model trained on your catalogue, and a multitrack arrangement you can finish elsewhere. None of those three pieces existed twelve months ago.

What I would actually use it for in May 2026

  • Demos and pre-production. v5.5 is fast enough and good enough that you can generate a full-band reference for a song idea before you commit to recording it. The Voices feature means the demo singer can be you.
  • Stem libraries for hybrid workflows. Generate, export, drop into a hardware or hybrid setup. The output is a starting point, not the finished part.
  • Custom model as a sound signature. If you release under an artist name, training a custom model on your catalogue is the closest thing to a portable production identity that the platform has shipped.

What I would not use it for

  • Releases that need clean copyright provenance. Even on a paid plan, the ownership wording is no longer “you own this.” If a sync agency or label is involved, that ambiguity matters. Until the licensed models ship and the documentation catches up, treat output as collaboration rather than authorship.
  • Highly specific creative briefs. v5.5 is significantly better at generic prompts than v5 was, but the same complaint that applied at launch still applies: very narrow stylistic asks still come back close-but-not-quite. This is consistent across the wider AI music landscape.

Where this is heading

The next milestone is the post-Warner licensed model, due later in 2026. When that ships, the existing model line is deprecated, the legal posture cleans up considerably, and the artist-opt-in system goes live in earnest. Until then, May 2026 is a useful window: v5.5 is mature, the personalisation features are real, and the platform is still operating under terms that make casual experimentation easy.