Suno API: 11 Endpoints, V5.5 Fenix, and Pay-Per-Use Pricing

Suno is no longer just a text-to-song tool. It is a complete AI audio platform, and music generation is one of eleven endpoints. The latest Chirp model, V5.5 Fenix, was released on March 26, 2026 and adds custom voice training, custom model fine-tuning, and stronger vocal expressiveness. On Unifically you can call all eleven Suno endpoints through a single async REST API with pay-per-generation pricing.

TL;DR: Suno on Unifically exposes 11 endpoints (Music, Extend, Cover, Add Vocals, Add Instrumental, Stems, Lyrics, WAV Export, Sound Effects, Voice, Custom Model). The current Chirp lineup is v3.5, v4, v4.5 Auk, v4.5 Plus Bluejay, v5.0 Crow, and v5.5 Fenix, selected via the mv parameter. Sound Effects is the cheapest call at $0.01. Custom Model training is $0.64 once and reusable indefinitely.

What is Suno today?

Suno is an AI audio platform built on the Chirp model family. You can describe a song in plain language and get audio back, hand Suno an instrumental and ask for a vocal pass, point it at one of your tracks and get clean stems back, generate sound effects with a target BPM and key, or train a custom voice that you reuse across all your future generations.

The current Chirp lineup includes Chirp v3.5, v4, v4.5 Auk, v4.5 Plus Bluejay, v5.0 Crow, and the newest v5.5 Fenix, selected via the mv parameter (v3.5, V4, V4_5, V4_5PLUS, V5, V5_5). Newer variants accept longer prompts and richer style tags than v3.5 and v4.

On Unifically every Suno tool runs on the same async pattern: submit a JSON request, then poll the shared Suno feed endpoint until the job returns audio (or text, in the case of Lyrics).

What V5.5 Fenix added

V5.5 Fenix landed on March 26, 2026 and added three things that previously required separate models:

Voices. Upload a singing voice sample and Suno trains a reusable voice profile with your exact vocal timbre.
Custom Models. Train a private Chirp checkpoint on 6 to 24 of your own tracks. Reuse the model across future Music and Cover calls.
My Taste. Personalised model that learns your style preferences over time.

The model itself improves on V5 in vocal expressiveness, voice fidelity, emotional depth, and prompt adherence. That matters for studios and brand teams that need a recognisable artist voice or a specific sonic identity.

What you can do with Suno

Every endpoint follows the same shape: POST a JSON request, receive a clip_id, poll until the job is complete. Below are all eleven endpoints with what they do, what they cost, and when to use them.

1. Music: full song generation

The flagship endpoint. Generate complete tracks with vocals, lyrics, and instrumentation.

Simple mode: one description prompt (gpt_description_prompt) and Suno handles the rest.
Custom mode: full lyric or idea prompt plus style tags and negative tags for fine creative control.
Optional song title and make_instrumental flag for backing-only output.
Pick any Chirp mv from v3.5 through V5.5 Fenix.
Price: $0.06 per generation.

2. Extend: continue a song from any timestamp

Take an existing Suno clip (by clip_id) or any uploaded audio (by audio_url) and continue it from a chosen continue_at time in seconds.

Same Simple / Custom split as Music.
Works with both Suno-generated and externally uploaded audio.
Price: $0.06 per generation.

3. Cover: AI cover versions in different styles

Reharmonise or restyle a track while keeping the melodic anchor. Provide a Suno clip_id or upload via audio_url, then describe the new genre or arrangement.

Optional make_instrumental for instrumental covers.
Chirp variants from v4 through V5.5 Fenix.
Price: $0.06 per generation.

4. Add Vocals: sing over your instrumental

Upload a backing track, point Add Vocals at its clip_id, and Suno generates a top line and lyric over it.

Optional start_s / end_s window in seconds to scope where vocals apply.
Plain-language vocal description via gpt_description_prompt, or a precise lyric via prompt.
Price: $0.06 per generation.

5. Add Instrumental: backing track for a vocal

The mirror of Add Vocals. Upload a vocal, describe the desired instrumentation, and Suno generates an arrangement underneath it.

Same start_s / end_s segment control.
Simple or Custom prompting.
Price: $0.06 per generation.

6. Extract Stems: vocals, drums, bass, guitar, keys

Pass a clip_id and get individual instrument stems back. Two modes:

Extract Stems (separate_vocal). Vocals + instrumental, 2 stems. $0.05 per generation.
Extract All Stems (split_stem). Full multi-track split (vocals, backing vocals, drums, bass, guitar, keys, percussion, strings, synth, fx, brass, woodwinds, depending on the source mix). $0.20 per generation.

7. Lyrics: text-only lyric generation

The cheapest full-cycle Suno endpoint. POST a prompt and choose mv. The response always returns two lyric variations (a and b) so you can pick or merge.

Pairs naturally with Music Custom mode. Paste a chosen lyric and render it as audio.
Price: $0.012 per generation.

8. WAV Export: lossless audio for your DAW

Take any Suno clip_id and export it as a WAV file suitable for mastering, archival, post production, and sample work.

Single-purpose call. Only clip_id required.
Price: $0.05 per generation.

9. Sound Effects: one-shots and tempo-locked loops

A dedicated SFX endpoint, not music. Describe the sound (e.g. "dog barking in street"), pick type: one-shot or loop, optionally lock to a bpm (1 to 300) and a musical key, and Suno returns the asset.

All Chirp variants from v3.5 through V5.5 Fenix supported.
Optional negative tags to exclude unwanted styles.
Price: $0.01 per generation. The platform's cheapest call.

10. Voice: train your own AI singer

Upload a voice recording (audio_url, up to 100 MB) plus a verification clip (verification_audio_url) and matching phrase_id from the verification-phrase endpoint, and Suno trains a reusable voice profile.

Set a singer skill level: Beginner, Intermediate, Advanced, or Professional.
Tag genre styles to bias the voice toward specific musical contexts.
Optional vocal_start_s / vocal_end_s to point Suno at the cleanest section of your source.
The trained voice ID can be reused across other Suno generations.
Price: $0.06 per generation.

11. Custom Model: fine-tune a private Suno checkpoint

For artists, brands, and studios that want a model that sounds like them. Upload between 6 and 24 reference tracks (each up to 100 MB), give the checkpoint a name, and Suno trains a custom model in an async job that polls every 30 seconds.

Returns a model identifier you can pass into future Music or Cover requests.
No prompt knobs. Training is entirely data-driven.
Price: $0.64 per training run (one-time, then reuse the model indefinitely).

Suno pricing on Unifically at a glance

All prices are per generation, transparent, and pay-as-you-go, with no subscriptions.

Endpoint	Price
Sound Effects	$0.01
Lyrics	$0.012
Extract Stems	$0.05
WAV Export	$0.05
Music	$0.06
Extend	$0.06
Cover	$0.06
Add Vocals	$0.06
Add Instrumental	$0.06
Voice	$0.06
Extract All Stems	$0.20
Custom Model (training)	$0.64 (one-time)

How to call Suno on Unifically

const API = 'https://api.unifically.com';
const headers = {
  Authorization: `Bearer ${process.env.UNIFICALLY_API_KEY}`,
  'Content-Type': 'application/json',
};

const start = await fetch(`${API}/suno/generate`, {
  method: 'POST',
  headers,
  body: JSON.stringify({
    mv: 'V5_5',
    gpt_description_prompt: 'A cinematic indie folk track, fingerpicked acoustic guitar, soft female vocals, building strings in the chorus',
  }),
}).then((r) => r.json());

while (true) {
  await new Promise((r) => setTimeout(r, 5000));
  const task = await fetch(`${API}/suno/feed/${start.clip_id}`, { headers }).then((r) => r.json());
  if (task.status === 'complete') {
    console.log(task.audio_url, task.lyrics);
    break;
  }
  if (task.status === 'failed') throw new Error(task.error);
}

Switch the path to /suno/extend, /suno/cover, /suno/stems, /suno/voice, etc. for the other endpoints. Each one keeps the same async pattern.

Workflows you can build now

Game audio pipelines

Generate one-shots for hits, pickups, and UI cues with Sound Effects. Use tempo-locked loops for combat or exploration beds. Train a Custom Model on your soundtrack to keep new music in the same sonic world.

Branded artist personas

Train a Voice profile once, then reuse the voice ID across Music, Cover, and Add Vocals. Every track keeps a consistent vocal identity. Useful for content series, virtual artists, and brand campaigns.

Studio-grade post production

Generate or extend a track, run Stems if you need to swap parts, then WAV Export for lossless handoff to mixing or mastering engineers. The whole loop stays in one API.

Lyrics-first content workflows

Use Lyrics to brainstorm two variations cheaply, lock the best lines, then drop them into Music Custom mode. You only pay $0.012 per lyric pass, far cheaper than rendering full audio for each iteration.

Private model for a studio or label

Custom Model lets a studio train a checkpoint on its back catalog or brand identity tracks for $0.64 once, then call it on demand from inside their own product. The trained model becomes a private generator that mirrors the studio's signature.

Things to know

Defaulting to V5.5 Fenix on every call. Older variants accept shorter prompts and bias toward different vocal characters. For stylised retro tracks, v4 or v4.5 Auk can be the better fit.
Calling Extract All Stems by default. It is 4x the price of Extract Stems. Use the cheaper separate_vocal mode unless you genuinely need the multi-track split.
Sending a non-MP3 audio file to Cover or Extend. Suno accepts standard streaming audio formats. Uploads outside the documented list will fail before the model runs.
Forgetting Voice needs a verification clip. Voice training requires both a singing sample and a matching verification clip with the right phrase_id. Without both, training silently fails.
Treating Custom Model as a per-call cost. It is one-time at $0.64. After training, reuse the model identifier indefinitely in Music and Cover requests.

Frequently asked questions

What is Suno V5.5 Fenix?

V5.5 Fenix is Suno's Chirp model released March 26, 2026. It improves vocal expressiveness, voice fidelity, and prompt adherence over V5 Crow, and adds Voice training, Custom Model fine-tuning, and the My Taste personalisation feature.

How many Suno endpoints does Unifically expose?

Eleven: Music, Extend, Cover, Add Vocals, Add Instrumental, Extract Stems, Lyrics, WAV Export, Sound Effects, Voice, and Custom Model. Each one has its own playground tab and its own per-generation rate.

What is the cheapest Suno call?

Sound Effects at $0.01 per generation. Lyrics is the next cheapest at $0.012, and is the cheapest call that produces a full creative output (two lyric variations per call).

How do I select a Chirp model version?

Pass the mv parameter on every generation request. Accepted values are v3.5, V4, V4_5 (Auk), V4_5PLUS (Bluejay), V5 (Crow), and V5_5 (Fenix). Newer values accept longer prompts and richer style tags.

Can I train a private Suno model?

Yes. Custom Model accepts 6 to 24 of your own reference tracks (up to 100 MB each), trains async, and returns a model identifier you can reuse in Music and Cover calls. Training is $0.64 once, then unlimited reuse.

Suno model page: live playgrounds for all 11 endpoints.
Suno Sound Effects and Suno Voice: the new V5.5 endpoints.
ElevenLabs: another option for TTS, dialogue, and voice changing.
Veo 3.1 vs SeeDance 2.0: pair Suno audio with the right video model.

Suno API: 11 Endpoints, V5.5 Fenix, and Pay-Per-Use Pricing

What is Suno today?

What V5.5 Fenix added