Suno Custom Models vs Custom Voices and Voice Clone
Suno Custom Models are private, account-bound and capped at 3. Here is what they are, what they need, and how they differ from custom voices and voice clone.
Let's settle one thing first: Suno does have an official "Custom Models" feature, but it is not the same thing as the "custom voices" or "voice clone" you see on third-party platforms. The former is a private feature inside your official account that a third-party API cannot reach; the latter is a substitute capability the third-party platform provides itself. This article treats them separately.
What the official Custom Models feature is
Everything below is stated publicly on the official Custom Models page (we checked it on 2026-09-26):
| Official wording | What it means |
|---|---|
| Fine-tune a model that matches your taste using your own uploads | It is trained, not picked from presets |
| You can create up to 3 custom models | There is a hard cap on the number |
| For your own use only | Not published publicly, and it does not enter the public catalog |
| You must own the rights to the audio used for training, including your own Suno library | The material has to be something you have the right to use |
| For safety and personalization, no one but you can access your custom models | A private resource tied to the account |
The official page also lists five common questions (what it is, what it needs, how many you can have, who can use it, whether V6 is supported), plus a Custom Models guide. For the exact rules and current plan terms, refer to the official page; we do not interpret it on their behalf.
Three names that get mixed up
| Name | What it is | Where it lives |
|---|---|---|
| Custom Models | A dedicated model fine-tuned on your own audio, officially capped at 3 | Your official account, usable only by you |
| Persona Voices | A voice feature in the official plans, listed as one of the benefits in the official paid plan description | Your official account |
| Custom voice / voice clone | Train a reusable singing voice from a recording, then assign it to sing when generating a song | Inside the third-party platform's account system |
Many people searching for "custom models" actually want the third kind of result - their own voice or their own style showing up in the song. A "custom voice" covers that need, without the official training entry point.
Why a third-party API cannot offer the official Custom Models
Two reasons stack up, and the mechanism simply does not work:
- The official side has no open platform. No keys are issued externally and no public API documentation exists, so a third party has no "officially authorized" way to call this feature.
- It is a private resource inside the account. A custom model is bound to the account that created it, while a third-party API uses the platform's own account pool, not your personal account. Your model does not travel with an API key.
So if someone tells you "our API can call your Suno custom model", that claim does not hold up mechanically.
What a third party can do: custom voices (voice clone)
Taking our platform as an example, here is the part that can be replaced:
| Step | Details |
|---|---|
| Creation | Upload or record a source voice, name it and follow the wizard; there is a second verification in the middle to confirm the voice belongs to you |
| Billing | Voice verification, cloning, status checks, deletion and download are free; generating a song with a cloned voice costs 1.2 yuan per run, twice the 0.6 yuan of a normal generation, and 2.4 yuan per run with Max Mode on |
| Supported combinations | Clone-only generation, cover plus clone, extend plus clone, inspiration plus clone |
| Cannot be combined | Trim, speed change, reverse, fusion, sampling, replace, add track and stem separation modes show as locked |
| Watch out for | A voice can expire. Check availability before each use; see the voice clone API for the exact endpoints |
The full walkthrough with screenshots is part 4 of the studio tutorial; to work directly through the API, see the voice clone and designated-voice generation API.
How to choose
| What you want | Which route to take |
|---|---|
| My own voice singing my songs | Voice clone / custom voice, which a third-party platform can do |
| Fine-tune a style and formula into a model that is only mine, used only inside my official account | Official Custom Models, which a third party cannot provide |
| Wire generation into my own product and produce songs at scale | A third-party API, starting with requesting a key |
FAQ
Do Custom Models cost extra?
The official page does not state pricing publicly, so we will not guess. Plans and benefits are governed by the official page and your own account dashboard.
Do the official Custom Models support V6?
The official FAQ does include this question, but the answer requires expanding the page to see. Refer to the official wording; we do not paraphrase it.
Can someone else use the voice I cloned?
The voice lives in your platform account and is only assigned when called with your account or a key you issued. No one else can reach it.
When I generate through a third-party API, can I specify a voice from my official account?
No. The two systems are separate: resources in your official account are only available inside that account, while a third-party API uses platform-side accounts.
Summary
In one line: the official Custom Models are a private training feature inside your account, capped at 3, out of reach for third parties; a third-party "custom voice" solves "make the song sing in your voice", and that we can do today. To hear the result first, generate a song with a cloned voice in the studio; to wire it into your own product, start by getting a key.
¥0.6 per generation, 2 tracks, unlimited downloads
Runs on the Suno V6 engine. Download MP3 or WAV as often as you like. Signing up needs an email address, not a phone number.
Start creating Read the API docs