List Music Models

GET /v1/music/models No API key needed View .md

The music models currently served, each with what it honours, so a client can adapt without code changes. false or [] means unsupported. Requests are checked against these entries before they are queued. A model without a valid capability spec is not listed.

Guides: Write songs for MiniMax Music 3, Make songs with ACE-Step 1.5

Responses

application/json · music-models-payload

1 field

data array of objects
30 fields
- id string
- name string
- canonical_slug string
- hugging_face_id string or null
- description string
- created integer
- output_modalities array of strings
- interaction string
  
  job: submit and poll.
- tasks array of objects
  
  What the model does, each with the audio it takes, the settings it reads, the regions it works on, the quality tiers it runs on and how long its song is.
  6 fields
  - task string
    
    The task value that asks for it, such as text2music.
  - requires array of objects
    
    The audio the task takes, by role: {role, min, max, min_seconds, max_seconds}. source is sent as source_audio, reference as reference_audio; a role it does not list is a 400.
  - parameters array of strings
    
    The model settings only this task reads (under parameters in this list); another task refuses them.
  - regions object or null
    
    {min, max}: how many regions ([{start, end}] in seconds of the source) the task takes; null where it takes none.
    
    2 fields
    
    min integer
    
    max integer
  - length string
    
    How long the song is. duration the length asked for; source the source's own (duration is refused); extend duration, the new total, longer than the source, billing only the seconds added.
    
    durationsourceextend
  - tiers array of strings
    
    The quality tiers the task runs on, the one it gets without quality first; empty for a model without tiers. Any other tier is a 400.
- duration object
  
  {semantics, min, max, default} in seconds; semantics is exact (the song is that long) or max (the song may end sooner).
- lyrics string
  
  requiredoptional
- lyrics_max_length integer or null
  
  The most characters lyrics may have.
- lyrics_tags object
  
  The section tags the model card spells (tags), and other spellings written its way (aliases, e.g. Inst as Instrumental); a bracketed tag that matches neither is left as written.
  2 fields
  - tags array of strings
  - aliases object
- best_of object or null
  
  The most takes best_of may ask for (max); null for a model that does not take best_of.
  1 field
  - max integer
- write object or null
  
  Whether the model takes write: true, and the languages its writer writes in (languages); null for a model that does not.
  1 field
  - languages array of strings
- instrumental boolean
  
  Whether the model honours instrumental set to true, a song without words. Such a model is sent its own instrumental tag ([Instrumental]) for a new song with no lyrics.
- prompt_max_length integer
- text_budget object or null
  
  The tokens prompt and lyrics may come to together, counted at chars_per_token characters a token; null where the model sets none.
  2 fields
  - max_tokens integer
  - chars_per_token number
- controls object
  
  bpm, key, time_signature the model takes, each {mode}: native (enforced) or advisory (a hint written into the caption). min and max bound bpm; values lists every value a control takes where the model takes only some (ace-step-1.5's time_signature: 2/4, 3/4, 4/4, 6/8).
- outputs object
  
  max_n, kinds, roles, formats, default_format, sample_rates, channels.
- seed boolean
- adapters boolean
- timestamps boolean
  
  Whether the model honours timestamps set to true, which times the lyrics of each song it makes.
- streaming object
  
  progressive: a job on this model has a stream_url to hear the song as it renders.
  1 field
  - progressive boolean
- allowed_passthrough_parameters array of strings
- parameters object
- attribution string or null
  
  The name the model's license asks to be shown beside it.
- languages object or null
  
  The languages the model sings lyrics in (values) and the one it assumes (default); null for a model that takes no language.
  2 fields
  - values array of strings
  - default string
- quality_tiers array of objects or null
  
  What quality buys, cheapest first; null for a model without tiers. A request that names no tier gets the one marked default. Each tier bills multiplier times the per-second price, which cost_usd states (null for an unpriced model).
  5 fields
  - name string required
  - aliases array of strings required
  - multiplier number required
  - default boolean required
  - cost_usd string or null required
- pricing_skus object
  
  {generate_per_second} in dollars per second of audio produced at the default quality tier; {} for an unpriced model.

curl https://api.vultrinference.com/v1/music/models

import requests

response = requests.get(
    "https://api.vultrinference.com/v1/music/models",
)
print(response.json())

const response = await fetch("https://api.vultrinference.com/v1/music/models");
console.log(await response.json());

Response · 200

{
    "data": [
        {
            "id": "string",
            "name": "string",
            "canonical_slug": "string",
            "hugging_face_id": "string",
            "description": "string",
            "created": 0,
            "output_modalities": [
                null
            ],
            "interaction": "string",
            "tasks": [
                {}
            ],
            "duration": {},
            "lyrics": "required",
            "lyrics_max_length": 0,
            "lyrics_tags": {
                "tags": [],
                "aliases": {}
            },
            "best_of": {
                "max": null
            },
            "write": {
                "languages": []
            },
            "instrumental": false,
            "prompt_max_length": 0,
            "text_budget": {
                "max_tokens": null,
                "chars_per_token": null
            },
            "controls": {},
            "outputs": {},
            "seed": false,
            "adapters": false,
            "timestamps": false,
            "streaming": {
                "progressive": null
            },
            "allowed_passthrough_parameters": [
                null
            ],
            "parameters": {},
            "attribution": "string",
            "languages": {
                "values": [],
                "default": null
            },
            "quality_tiers": [
                {}
            ],
            "pricing_skus": {}
        }
    ]
}