> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mayaresearch.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Models

> Three models, and which transports each one answers on.

Pass `model` on `start` (WebSocket) or in the body (HTTP). Omit it for the
default.

| `model`                       | WebSocket | Notes                                                     |
| :---------------------------- | :-------- | :-------------------------------------------------------- |
| *(omitted)* — `Maya 2 Native` | yes       | The default. Region-local and pre-warmed.                 |
| `Maya 2 Native Emotional`     | yes       | Emotion tags inline in `text`, e.g. `"[happy] नमस्ते!"`   |
| `Maya 2 Global`               | **no**    | HTTP only — `POST /v1/tts` with `"model":"Maya 2 Global"` |

<Warning>
  On a socket, `model` is **sticky for the connection** and cannot be changed
  afterwards. Open a new socket to switch.
</Warning>

## Maya 2 Native Emotional

Runs as a single deployment in Mumbai and is **not pre-warmed**, so a US caller
pays the crossing and the first turn on a connection pays socket setup.

Everything else is identical — same URL, same key, same frames, same 24 kHz.

## Maya 2 Global

Not available on the WebSocket at all. Requesting it there is an error rather
than a silent fallback, so you cannot accidentally be served by a different
model than you asked for.

```json theme={null}
{ "text": "...", "model": "Maya 2 Global" }
```