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

# Model configuration

> Voice agents rely on AI models: LLM (understanding and responses), TTS (voice) and STT (speech transcription). By default they run on models managed by Brosbots.

## Brosbots-managed models (default)

After signing up at [app.brosbots.com](https://app.brosbots.com) your
agents run on the platform-managed model stack right away - no
configuration and no keys needed. Calls are billed from your balance at
your plan's per-minute rate.

## Where models are configured

* **Voice (TTS)** and **transcription (STT)** are set **per agent** - in
  the agent editor, in the **Voice** and **Transcription** tabs. Each
  agent can speak with a different voice and have different speech
  recognition settings.
* **Your own provider keys (BYOK)** are added once for the whole
  organization in **Account → Provider keys**. Once a key is added, that
  provider becomes available in the selection lists in agents.

## Your own providers (BYOK) - by plan

| Plan | Models |
| - | - |
| **Free** | Brosbots-managed stack (no custom keys) |
| **Pro** | BYOK: your own OpenAI, Groq, Google, Deepgram keys |
| **Business** | All providers (incl. ElevenLabs, Cartesia and Soniox for voice, Soniox and Speechmatics for transcription) + custom voices |
| **Enterprise** | Custom terms and SLA |

The following sections describe each model type in detail.


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