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

# AI Model Settings

> Configure AI providers and per-purpose model overrides in the Web Console.

## Overview

The Models settings page lets you configure which AI providers and models your tenant uses. Connect multiple providers and optionally override the model used for specific tasks.

## Provider configuration

Configure one or more AI providers:

| Provider              | Required fields                          |
| --------------------- | ---------------------------------------- |
| **Anthropic**         | API key                                  |
| **OpenAI**            | API key                                  |
| **Gemini**            | API key                                  |
| **Azure OpenAI**      | Resource name, API key, API version      |
| **Amazon Bedrock**    | Region, access key ID, secret access key |
| **Google Vertex**     | Project ID, location                     |
| **OpenAI-compatible** | Base URL, API key                        |

### Provider status

| Status         | Meaning                                     |
| -------------- | ------------------------------------------- |
| Connected      | All required fields configured and tested   |
| Pending        | Partial configuration (some fields missing) |
| Not configured | No fields set                               |

### Testing a connection

After entering credentials, click **Test Connection** to verify the provider is reachable and the credentials are valid.

## Model selection

For each configured provider, select the default model ID to use (e.g., `claude-sonnet-4-6`, `gpt-4o`).

## Per-purpose model overrides

With Business+ plan, you can override the model used for specific tasks:

| Purpose             | Description                          | Default    |
| ------------------- | ------------------------------------ | ---------- |
| `exploration`       | Runbook generation                   | Main model |
| `exploration-light` | Routine exploration steps            | Main model |
| `selector`          | Selector resolution (high frequency) | Main model |
| `extraction`        | Data extraction (high frequency)     | Main model |
| `review`            | Review and fix suggestions           | Main model |
| `fallback`          | Agent Fallback                       | Main model |
| `vision`            | Vision Fallback (screenshot-based)   | Main model |

<Tip>
  Use a faster, cheaper model (e.g., `claude-haiku-4-5-20251001`) for high-frequency tasks like `selector` and `extraction` to reduce cost and latency.
</Tip>