> ## 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.

# AWS Bedrock LLM for Bolna voice agents

> Run OpenAI's gpt-5.6-luna on AWS Bedrock in India as the LLM of a Bolna voice agent, so LLM requests and data stay in India.

[AWS Bedrock](https://aws.amazon.com/bedrock/) serves OpenAI's **gpt-5.6-luna** from AWS's India region. Bolna calls it through Bedrock's India inference profile, so every LLM request and its data stay in India. Use it when you need gpt-5.6-luna with data residency in India.

***

## Quick config

In the dashboard, open the agent's [Intelligence tab](https://www.bolna.ai/docs/agent-setup/llm-tab.md), pick **bedrock** as the provider and **gpt-5.6-luna** as the model. Over the API:

```json theme={"system"}
"llm_agent": {
  "agent_type": "simple_llm_agent",
  "agent_flow_type": "streaming",
  "llm_config": {
    "provider": "bedrock",
    "model": "bedrock/gpt-5.6-luna",
    "max_tokens": 150,
    "temperature": 1,
    "reasoning_effort": "none"
  }
}
```

Bedrock runs on Bolna's AWS account. There is no key to connect.

***

## Supported models

| Model | Shown as | Served from | Best for |
| - | - | - | - |
| `bedrock/gpt-5.6-luna` | gpt-5.6-luna | AWS India (Mumbai and Hyderabad) | Low-cost voice agents that need data residency in India |

***

## Key settings

| Setting | Type | Recommended | Description |
| - | - | - | - |
| `provider` | string | `"bedrock"` | Provider name |
| `model` | string | `"bedrock/gpt-5.6-luna"` | Model to use |
| `max_tokens` | integer | `150` | Cap on response length; keep it short for voice |
| `temperature` | float | `1` | gpt-5.6-luna is a reasoning model and only accepts `1` |
| `reasoning_effort` | string | `"none"` | How much the model reasons before answering: `none`, `low`, `medium`, `high` or `xhigh`. Higher values add latency |
| `agent_flow_type` | string | `"streaming"` | Always `"streaming"` for voice |

***

## Things to plan for

* **Indian servers.** A Bedrock agent can run its calls on Bolna's Indian servers when the rest of its setup qualifies. See [Routing calls through Indian servers](https://www.bolna.ai/docs/enterprise/indian-server-configuration.md).
* **Twilio calls run on US servers.** The LLM still runs in India, but each reply crosses from the US to India, which adds latency.
* **Graph agents.** A node can switch to Bedrock through its own `llm_config`.

***

## Related

* [Intelligence Tab](https://www.bolna.ai/docs/agent-setup/llm-tab.md): configure the LLM in the dashboard
* [OpenAI](https://www.bolna.ai/docs/openai-llm.md): gpt-5.6-luna and the rest of the GPT-5 line, served by OpenAI
* [Routing calls through Indian servers](https://www.bolna.ai/docs/enterprise/indian-server-configuration.md): keep the whole call in India
* [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.