API reference

Voices and models

Browse the voice and model catalogs, register custom language models, and estimate cost per minute.

These routes describe what you can put into an agent configuration: text-to-speech providers, their models and voices, the language model and speech-to-text catalogs, and your own OpenAI-compatible models. One route estimates the per-minute cost of a combination.

All routes need Authorization: Bearer <key>.

Note: The catalogs list what an agent configuration may name. Whether a given third-party provider actually works on live calls depends on the credentials and models the operator has set up for the voice engine. When the operator has configured the platform's local models, they appear first in each catalog with the provider local. A local language model is listed with "base_url": null and "hosted_by_platform": true: where it runs is not part of the catalog, and no key or address is needed to use it.

Operation	Method and path
List TTS providers for a language	`GET /api/v1/voice-config/tts`
List voices	`GET /api/v1/voice-config/tts/voices`
List languages	`GET /api/v1/voice-config/languages`
List all voices and models	`GET /user/model/all`
Add a custom model	`POST /user/model/custom`
List custom models	`GET /user/model/custom`
Remove a custom model	`DELETE /user/model/custom/{model_id}`
Estimate the price per minute	`GET /get_current_price`

Operations

List TTS providers for a language

GET /api/v1/voice-config/tts

Returns the text-to-speech providers that have at least one model for a language, with those models.

Auth: API key or session token. Role: viewer and up.

Parameter	In	Type	Required	Description
`language`	query	string	yes	Language code, for example `en` or `hi`.

Request

bash

curl "https://api.sevennadi.si/api/v1/voice-config/tts?language=en" \
  -H "Authorization: Bearer vk-0123456789abcdef0123456789abcdef"

Response 200 OK

json

{
  "language": "en",
  "providers": [
    {
      "id": "d8b6e33f-f5b4-5e99-bd65-25c34a93d1df",
      "name": "ElevenLabs",
      "slug": "elevenlabs",
      "is_supported": true,
      "models": [
        {
          "id": "2c1127ca-1ba7-5975-9257-6f4321c0d7df",
          "model_id": "eleven_flash_v2_5",
          "display_name": "Eleven Flash v2.5",
          "description": "Ultra-low latency model in 32 languages.",
          "display_order": 1,
          "quality_tier": "flash",
          "is_supported": true,
          "default": false
        }
      ]
    }
  ],
  "custom_voices": {},
  "default": null
}

Errors

Status	Message
400	`query.language: Field required`

List voices

GET /api/v1/voice-config/tts/voices

Returns the voices of one provider and model, optionally narrowed to a language.

Auth: API key or session token. Role: viewer and up.

Parameter	In	Type	Required	Description
`provider_id`	query	string	yes	The provider's `id` or its `slug`.
`model_id`	query	string	yes	The model's `model_id`, for example `eleven_flash_v2_5`.
`language`	query	string	no	Language code.
`page`	query	integer	no	Starts at 1. Default `1`.
`page_size`	query	integer	no	1 to 200. Default `50`.

Request

bash

curl "https://api.sevennadi.si/api/v1/voice-config/tts/voices?provider_id=elevenlabs&model_id=eleven_flash_v2_5&language=hi" \
  -H "Authorization: Bearer vk-0123456789abcdef0123456789abcdef"

Response 200 OK

json

{
  "items": [
    {
      "id": "50cb8479-fb6e-5523-a54b-a5018a6a2d5a",
      "provider_id": "d8b6e33f-f5b4-5e99-bd65-25c34a93d1df",
      "voice_id": "iWNf11sz1GrUE4ppxTOL",
      "name": "Viraj - Warm, Energetic and Lively",
      "gender": "male",
      "accent": "standard",
      "age_range": "young",
      "use_case": ["conversational"],
      "is_native": true,
      "preview_url": null,
      "source": "platform"
    }
  ]
}

The response has no total; request the next page until items comes back shorter than page_size. Use voice_id and name in the agent's synthesizer.provider_config.

Errors

Status	Message
400	`query.provider_id: Field required` or `query.model_id: Field required`
404	`Provider not found`

List languages

GET /api/v1/voice-config/languages

Returns the language codes the catalogs use. This route is not part of the core OpenAPI document.

Auth: API key or session token. Role: viewer and up.

No parameters.

Request

bash

curl https://api.sevennadi.si/api/v1/voice-config/languages \
  -H "Authorization: Bearer vk-0123456789abcdef0123456789abcdef"

Response 200 OK

json

[
  {"code": "hi", "name": "Hindi", "iso_639_1": "hi"},
  {"code": "en", "name": "English", "iso_639_1": "en"}
]

List all voices and models

GET /user/model/all

Returns three catalogs in one response: voices, language models (including your custom models) and speech-to-text models. This route is not part of the core OpenAPI document.

Auth: API key or session token. Role: viewer and up.

No parameters.

Request

bash

curl https://api.sevennadi.si/user/model/all \
  -H "Authorization: Bearer vk-0123456789abcdef0123456789abcdef"

Response 200 OK

json

{
  "voices": [
    {
      "id": "50cb8479-fb6e-5523-a54b-a5018a6a2d5a",
      "voice_id": "iWNf11sz1GrUE4ppxTOL",
      "provider": "elevenlabs",
      "name": "Viraj - Warm, Energetic and Lively",
      "model": "eleven_flash_v2_5",
      "language_code": "hi",
      "accent": "Standard Male"
    }
  ],
  "llmModels": [
    {
      "id": "9acd3318-bd9b-57c2-a432-0471810ef198",
      "provider": "openai",
      "model": "gpt-4.1-mini",
      "display_name": "GPT-4.1 mini",
      "family": "openai",
      "deprecated": false,
      "custom": false,
      "base_url": null
    }
  ],
  "asrs": [
    {
      "id": "53eeb72f-a587-5db9-8886-97f4e9603470",
      "provider": "deepgram",
      "model": "nova-3",
      "name": "nova-3",
      "languages": ["en", "multi-hi", "es", "fr"],
      "supports_keywords": true,
      "supports_keyword_weights": false,
      "supports_context": false
    }
  ]
}

Your custom models appear in llmModels with "provider": "custom" and "custom": true. The lists above are shortened to one entry each.

Add a custom model

POST /user/model/custom

Registers an OpenAI-compatible language model endpoint that you host or rent. To use it, set the agent's llm_config.model to the model's name. The key is encrypted at rest and never returned. A sub-account also sees the custom models of its parent account.

Auth: API key or session token. Role: admin and up.

Parameter	In	Type	Required	Description
`custom_model_name`	body	string	yes	The model name. No other model you can see may have the same name.
`custom_model_url`	body	string	yes	Base URL, starting with `https://` or `http://`.
`custom_model_key`	body	string	yes	API key for the endpoint.
`custom_model_family`	body	string	yes	Model family, for your reference.
`custom_model_languages`	body	string	no	Languages the model handles, as one comma-separated string (at most 300 characters), for example `hi,en`.

Request

bash

curl -X POST https://api.sevennadi.si/user/model/custom \
  -H "Authorization: Bearer vk-0123456789abcdef0123456789abcdef" \
  -H "Content-Type: application/json" \
  -d '{
    "custom_model_name": "bank-assistant-v2",
    "custom_model_url": "https://llm.example.test/v1",
    "custom_model_key": "<your key>",
    "custom_model_family": "llama"
  }'

Response 200 OK

json

{"message": "model added successfully", "status": "added"}

Errors

Status	Message
400	`custom_model_name is required` (likewise for `custom_model_url`, `custom_model_key`, `custom_model_family`)
400	`custom_model_url must be an http(s) URL`
400	`A custom model with this name already exists`
403	`Your role (member) can't change custom model providers. It needs admin access; ask an owner or admin of the workspace.`

List custom models

GET /user/model/custom

Returns your custom models, oldest first, and those of the parent account if you are a sub-account.

Auth: API key or session token. Role: viewer and up.

No parameters.

Request

bash

curl https://api.sevennadi.si/user/model/custom \
  -H "Authorization: Bearer vk-0123456789abcdef0123456789abcdef"

Response 200 OK

json

{
  "models": [
    {
      "id": "1f2e3d4c-5b6a-4978-8a9b-0c1d2e3f4a5b",
      "model": "bank-assistant-v2",
      "display_name": "bank-assistant-v2",
      "provider": "custom",
      "family": "llama",
      "base_url": "https://llm.example.test/v1"
    }
  ]
}

Remove a custom model

DELETE /user/model/custom/{model_id}

Deletes one of your own custom models. A sub-account cannot delete a model that belongs to its parent.

Auth: API key or session token. Role: admin and up.

Parameter	In	Type	Required	Description
`model_id`	path	string (UUID)	yes	The model's `id`.

Request

bash

curl -X DELETE https://api.sevennadi.si/user/model/custom/1f2e3d4c-5b6a-4978-8a9b-0c1d2e3f4a5b \
  -H "Authorization: Bearer vk-0123456789abcdef0123456789abcdef"

Response 200 OK

json

{"message": "model deleted successfully", "status": "deleted"}

Errors

Status	Message
404	`Model not found`

Estimate the price per minute

GET /get_current_price

Estimates what one minute costs for a chosen speech-to-text, language model and text-to-speech, from your workspace's price book. It also says how many minutes the current wallet balance would cover. This is an estimate; each call is priced from its real usage when it ends. This route is not part of the core OpenAPI document.

Auth: API key or session token. Role: viewer and up.

Parameter	In	Type	Required	Description
`asr_provider`	query	string	no	Speech-to-text provider.
`asr_model`	query	string	no	Speech-to-text model.
`llm_provider`	query	string	no	Language model provider.
`llm_model`	query	string	no	Language model.
`tts_provider`	query	string	no	Text-to-speech provider.
`tts_model`	query	string	no	Text-to-speech model.

Request

bash

curl "https://api.sevennadi.si/get_current_price?asr_provider=deepgram&asr_model=nova-2&llm_model=gpt-4o-mini&tts_provider=elevenlabs" \
  -H "Authorization: Bearer vk-0123456789abcdef0123456789abcdef"

Response 200 OK

json

{
  "data": "0.094",
  "asr_cost": 0.008,
  "llm_cost": 0.006,
  "tts_cost": 0.05,
  "agent_cost": 0.064,
  "platform_cost": 0.02,
  "telephony_cost": 0.01,
  "total_cost": 0.094,
  "currency": "INR",
  "state": "success",
  "providers": {},
  "custom_price": false,
  "talk_time": 53
}

All costs are per minute in major units of currency. data is total_cost as a string. custom_price is true when your workspace has its own prices. talk_time is the wallet balance divided by total_cost, rounded down. The figures shown are an example of the shape; the real ones come from the operator's price book. See Billing.