DocumentationAPI ReferenceChangelog

Envocal API

API Reference Tools guide MCP server API Changelog

Endpoints

List models GETGet a model GETList voices GETGet a voice GET
Models & voices

Get a model

List models GET

The AI models the tenant's agents can use, in display order. An agent's model is one of these keys; a model's reasoningLevels are the values its reasoningLevel can take. Models marked preview can be chosen but may still change.

List voices GET

The voices the tenant's agents can speak with, in display order. An agent's voice is one of these keys, and it must work with the agent's model: pass model to get exactly the voices an agent on that model can use. Each voice has a short sample clip (sampleUrl).

GET
/v1/tenants/{tenantId}/models/{key}

Required permission: agents:read

Authorization

Bearer
headerAuthorizationBearer <token>

Tenant API key (evk_…), created in Studio → Vault → API keys.

Path Parameters

tenantId*string

Tenant id, or current for the key's own tenant. Must be the key's tenant or one of its sub-tenants.

key*string

Model key, e.g. gpt-realtime-1.5.

Response Body

The model.

  1. response
  1. response
  2. …

An AI model an agent can use.

key*string

The value to put in an agent's model, e.g. gpt-realtime-1.5.

name*string

Display name.

description?string

What the model is good at.

provider*

The company behind the model.

kind*string

Role in an agent: speech-to-speech models hear and speak themselves and come with their own voices; text models (LLMs) need separate speech recognition and voices; tts and stt are text-to-speech and speech-to-text models. Agents can use speech-to-speech models today.

Value in"speech-to-speech""text""tts""stt"
inputModalities*array<string>

What the model takes in: audio, text.

outputModalities*array<string>

What the model gives out: audio, text.

languages*array<string>

Every language the model speaks (BCP 47, e.g. sv, en-GB). Its multilingual voices speak these, unless a voice lists fewer. Empty when the provider doesn't state a list: then any language counts.

supportsTools?boolean

Whether the model can call tools.

reasoningLevels*array<string>

For reasoning models, the levels an agent's reasoningLevel can take, from fastest to most thorough. Empty for models that don't reason.

defaultReasoningLevel?string

The level used when an agent doesn't set one; null for models that don't reason.

isDefault?boolean

The model new agents get when none is given.

defaultVoice?string

The voice new agents get with this model when none is given.

status*string

preview (can be chosen, may change), active, or deprecated (still works, will be retired).

Value in"preview""active""deprecated"
Example Request
curl -X GET 'https://api.envocal.ai/v1/tenants/string/models/string'
Example Response
{  "key": "string",  "name": "string",  "description": "string",  "provider": {    "key": "string",    "name": "string",    "logoUrl": "string"  },  "kind": "speech-to-speech",  "inputModalities": [    "string"  ],  "outputModalities": [    "string"  ],  "languages": [    "string"  ],  "supportsTools": true,  "reasoningLevels": [    "string"  ],  "defaultReasoningLevel": "string",  "isDefault": true,  "defaultVoice": "string",  "status": "preview"}
GET
/v1/tenants/{tenantId}/models/{key}
Example Request
curl -X GET 'https://api.envocal.ai/v1/tenants/string/models/string'
{
  "key": "string",
  "name": "string",
  "description": "string",
  "provider": {
    "key": "string",
    "name": "string",
    "logoUrl": "string"
  },
  "kind": "speech-to-speech",
  "inputModalities": [
    "string"
  ],
  "outputModalities": [
    "string"
  ],
  "languages": [
    "string"
  ],
  "supportsTools": true,
  "reasoningLevels": [
    "string"
  ],
  "defaultReasoningLevel": "string",
  "isDefault": true,
  "defaultVoice": "string",
  "status": "preview"
}