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

# Sarvam LLM for Bolna voice agents

> Configure Sarvam 105B Chat (sarvam-105b-conversations), Sarvam's conversational LLM for Indian-language voice agents, as the LLM of a Bolna agent.

[Sarvam](https://www.sarvam.ai/) builds models for Indian languages. **Sarvam 105B Chat** (`sarvam-105b-conversations`) is the conversational variant of Sarvam 105B, tuned for real-time dialogue and voice agents. It handles native-script, romanized and code-mixed input in 10 Indian languages and English.

***

## Quick config

In the dashboard, open the agent's [Intelligence tab](https://www.bolna.ai/docs/agent-setup/llm-tab.md), pick **Sarvam** as the provider and **Sarvam 105B Chat** as the model. Over the API:

```json theme={"system"}
"llm_agent": {
  "agent_type": "simple_llm_agent",
  "agent_flow_type": "streaming",
  "llm_config": {
    "provider": "sarvam",
    "model": "sarvam-105b-conversations",
    "max_tokens": 150,
    "temperature": 0.2
  }
}
```

To use your own Sarvam API key, connect it at [platform.bolna.ai/auth/sarvam](https://platform.bolna.ai/auth/sarvam).

***

## Supported models

| Model | Shown as | Context | Best for |
| - | - | - | - |
| `sarvam-105b-conversations` | Sarvam 105B Chat | 32K tokens | Indian-language voice agents |

***

## Key settings

| Setting | Type | Recommended | Description |
| - | - | - | - |
| `provider` | string | `"sarvam"` | Provider name |
| `model` | string | `"sarvam-105b-conversations"` | Model to use |
| `max_tokens` | integer | `150` | Cap on response length; keep it short for voice |
| `temperature` | float | `0.2` | Lower is more consistent, good for scripted calls |
| `agent_flow_type` | string | `"streaming"` | Always `"streaming"` for voice |

`reasoning_effort` and `verbosity` don't apply to this model.

***

## Things to plan for

* **32K context window.** The limit covers the whole request: system prompt, conversation history, tool definitions and the reply (`max_tokens`). A request over the limit fails and the call ends, so keep long prompts and knowledge-heavy nodes well under it.
* **Ending calls.** With the `end_call` tool enabled, the model tends to end the call as soon as the task in the prompt looks done, sometimes before it confirms the details back to the caller. Test how your calls end before going live.
* **Graph agents.** Routing between nodes runs on Bolna's routing model; only the replies come from Sarvam. A node can also switch to Sarvam through its own `llm_config`.
* **Prompt caching.** Sarvam caches a repeated prompt prefix automatically, which speeds up later turns of a call. No configuration is needed.

***

## FAQ

> [accordion] Which languages does it support?
>
> Hindi, Bengali, Gujarati, Kannada, Malayalam, Marathi, Odia, Punjabi, Tamil, Telugu and English, in native script, romanized or code-mixed. Pair it with [Sarvam speech-to-text](https://www.bolna.ai/docs/sarvam-transcriber.md) and [Sarvam text-to-speech](https://www.bolna.ai/docs/sarvam-voice.md) for an all-Sarvam agent.

***

## Related

* [Intelligence Tab](https://www.bolna.ai/docs/agent-setup/llm-tab.md): configure the LLM in the dashboard
* [Sarvam text-to-speech](https://www.bolna.ai/docs/sarvam-voice.md) and [Sarvam speech-to-text](https://www.bolna.ai/docs/sarvam-transcriber.md)
* [Custom LLM inference](https://www.bolna.ai/docs/custom-llm-inference): bring your own OpenAI-compatible endpoint
* [Prompting Guide](https://www.bolna.ai/docs/prompting/introduction): write effective prompts for voice


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.