List available models

GET /api/v1/agents/models/available

Returns every model you can select for an agent. Each record carries the exact provider and id values to put in your agent config, so you can build model pickers, validate configs before saving, or keep your own systems in sync with the platform.

The list is built live from the platform's model catalog. When a model is added or removed, this endpoint reflects it immediately — poll it instead of hard-coding model ids.

Authentication

X-API-Key: zsk_...

Query parameters

Parameter	Type	Required	Description
`type`	`"llm"` \| `"stt"` \| `"tts"`	No	Only return models of this type. Omit to get all three.

Response

200 OK

{
  "object": "list",
  "data": [
    {
      "id": "gpt-4.1",
      "object": "model",
      "type": "llm",
      "provider": "openai",
      "name": "GPT-4.1",
      "is_default": false,
      "languages": null
    },
    {
      "id": "flux-general-multi",
      "object": "model",
      "type": "stt",
      "provider": "flux",
      "name": "Flux Multilingual",
      "is_default": false,
      "languages": ["en", "hi", "es", "fr", "ja", "de", "pt", "ru"]
    },
    {
      "id": "eleven_v4_turbo",
      "object": "model",
      "type": "tts",
      "provider": "elevenlabs",
      "name": "Eleven v4 Turbo",
      "is_default": true,
      "languages": ["en", "hi", "gu", "ta", "te", "es", "fr", "ja", "de", "zh", "ar", "pt", "ru", "mr", "bn", "kn", "ml", "pa", "ur", "id", "or", "ne"]
    }
  ]
}

List object

Field	Type	Description
`object`	`"list"`	Always `"list"`.
`data`	`Model[]`	Every matching model, ordered by `type` (`llm`, `stt`, `tts`), then `provider`, then `id`.

Model object

Field	Type	Description
`id`	`string`	The model id. Send it as the agent's `llm.model`, `stt.model` or `tts.model`.
`object`	`"model"`	Always `"model"`.
`type`	`"llm"` \| `"stt"` \| `"tts"`	Which part of the agent config the model belongs to.
`provider`	`string`	The provider id. Send it as the agent's `llm.provider`, `stt.provider` or `tts.provider`.
`name`	`string`	Human-readable model name, for display.
`is_default`	`boolean`	`true` for the model an agent uses when it sends `model: null` for that provider. Exactly one per STT and TTS provider. Always `false` for LLMs — an agent's `llm.model` is required.
`languages`	`string[]` \| `null`	Language codes the model supports, in the format of the agent's `languages` list. Every language in an agent's `languages` must be supported by both its STT and its TTS model. `null` for LLMs.

Use a model in an agent

Copy provider and id from a record into the matching section of the agent config:

{
  "llm": { "provider": "openai", "model": "gpt-4.1" },
  "stt": { "provider": "flux", "model": "flux-general-multi" },
  "tts": { "provider": "elevenlabs", "model": "eleven_v4_turbo" },
  "languages": ["hi", "en"]
}

Provider and model go together

A model id is always sent together with its provider. Picking a model from this list and sending only model is not enough — the agent API needs provider to know which service to call.

Errors

Status	When
`401`	Missing or invalid API key.
`400`	The API key's user has no organization selected.
`422`	`type` is not one of `llm`, `stt`, `tts`.
`503`	The model catalog is temporarily unavailable. Retry — the response is never an empty list.

Examples

curl "https://dashboard.zoxa.ai/api/v1/agents/models/available?type=tts" \
  -H "X-API-Key: zsk_..."

const { data } = await fetch(
  "https://dashboard.zoxa.ai/api/v1/agents/models/available",
  { headers: { "X-API-Key": "zsk_..." } },
).then((r) => r.json());

// TTS models that can speak Hindi
const hindiTts = data.filter((m) => m.type === "tts" && m.languages.includes("hi"));

import httpx

models = httpx.get(
    "https://dashboard.zoxa.ai/api/v1/agents/models/available",
    headers={"X-API-Key": "zsk_..."},
).json()["data"]

stt_defaults = {m["provider"]: m["id"] for m in models if m["type"] == "stt" and m["is_default"]}

Check the catalog from your terminal

Quick checks with curl and jq. Store your key in an environment variable first so it stays out of your shell history:

export ZOXA_KEY="zsk_..."

Count models per type

curl -sS https://dashboard.zoxa.ai/api/v1/agents/models/available \
  -H "X-API-Key: $ZOXA_KEY" | jq -r '.data | group_by(.type)[] | "\(.[0].type) \(length)"'

Prints one line per type, such as stt 6.

List every model as a table

curl -sS https://dashboard.zoxa.ai/api/v1/agents/models/available \
  -H "X-API-Key: $ZOXA_KEY" \
  | jq -r '.data[] | "\(.type)\t\(.provider)\t\(.id)\t\(.name)\tdefault=\(.is_default)"' \
  | column -t -s $'\t'

Find the default model for each STT and TTS provider

curl -sS https://dashboard.zoxa.ai/api/v1/agents/models/available \
  -H "X-API-Key: $ZOXA_KEY" | jq -r '.data[] | select(.is_default) | "\(.type) \(.provider) → \(.id)"'

Find models that support a language — here, Hindi TTS

curl -sS "https://dashboard.zoxa.ai/api/v1/agents/models/available?type=tts" \
  -H "X-API-Key: $ZOXA_KEY" | jq -r '.data[] | select(.languages | index("hi")) | "\(.provider) \(.id)"'

What a healthy response looks like

Every STT and TTS provider has exactly one model with is_default: true, every STT and TTS model has a non-empty languages list, and every LLM has languages: null.

POST /agents — the config these models plug into
Agent config schema — every field of llm, stt and tts
LLM providers · TTS providers · STT providers — latency, cost and quality per model

List available models

On this page