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

# Provider Authentication

> Configure AI providers using API keys or OAuth in Reliant

Reliant supports multiple AI model providers, allowing you to choose the best models for your workflow. Providers are configured through one of two authentication methods:

* **API key** (most providers)
* **OAuth login** (Claude Code and Codex)

You can configure API key providers through environment variables, configuration files, or the Settings UI. Claude Code and Codex authentication is completed in the Settings UI via their respective **Login** buttons.

## Supported Providers

Reliant supports the following AI model providers:

### Anthropic (Claude)

**Available Models**: Claude 4.5 Sonnet, Claude 4.6 Opus, Claude 4.5 Haiku

**Configuration**:

```bash theme={null}
export ANTHROPIC_API_KEY="sk-ant-..."
```

Or in `.reliant.json`:

```json theme={null}
{
  "providers": {
    "anthropic": {
      "apiKey": "sk-ant-...",
      "disabled": false
    }
  }
}
```

**Getting an API Key**:

* Sign up at [console.anthropic.com](https://console.anthropic.com)

### OpenAI

**Available Models**: GPT-5.3 Codex (flagship), GPT-5.2 Codex, GPT-5.2, GPT-5 Mini, GPT-5.2 Pro

**Configuration**:

```bash theme={null}
export OPENAI_API_KEY="sk-..."
```

Or in `.reliant.json`:

```json theme={null}
{
  "providers": {
    "openai": {
      "apiKey": "sk-...",
      "disabled": false
    }
  }
}
```

**Getting an API Key**:

* Sign up at [platform.openai.com](https://platform.openai.com)
* Navigate to API Keys section
* Create new secret key

### Claude Code (OAuth)

**Available Models**: Claude 4.5 Sonnet, Claude 4.6 Opus, Claude 4.5 Haiku

**Authentication Method**: OAuth (no API key required — uses your Claude account)

**How to Connect**:

1. Open **Settings → AI**
2. Select **Claude Code**
3. Click **Login with Claude Code**
4. Complete sign-in in your browser
5. Return to Reliant and verify the success state

<Note>
  If you're using the **web UI** (not the desktop app), see [OAuth in Web Mode](#oauth-in-web-mode) below.
</Note>

### Codex (OAuth)

**Available Models**: GPT-5.3 Codex, GPT-5.3 Codex Spark, GPT-5.2 Codex

**Authentication Method**: OAuth (no API key required)

**How to Connect**:

1. Open **Settings → AI**
2. Select **Codex**
3. Click **Login with Codex**
4. Complete sign-in in your browser
5. Return to Reliant and verify the success state

<Note>
  If you're using the **web UI** (not the desktop app), see [OAuth in Web Mode](#oauth-in-web-mode) below.
</Note>

### Google Gemini

**Available Models**: Gemini 3.1 Pro, Gemini 3.1 Pro (Custom Tools), Gemini 3 Pro, Gemini 3 Flash, Gemini 2.5 Pro, Gemini 2.5 Flash

**Configuration**:

```bash theme={null}
export GEMINI_API_KEY="..."
```

Or in `.reliant.json`:

```json theme={null}
{
  "providers": {
    "gemini": {
      "apiKey": "...",
      "disabled": false
    }
  }
}
```

**Getting an API Key**:

* Visit [ai.google.dev](https://ai.google.dev)
* Sign in with Google account
* Create API key in API Keys section

### OpenRouter

**Configuration**:

```bash theme={null}
export OPENROUTER_API_KEY="sk-or-..."
```

Or in `.reliant.json`:

```json theme={null}
{
  "providers": {
    "openrouter": {
      "apiKey": "sk-or-...",
      "disabled": false
    }
  }
}
```

**Note on Model Support**: OpenRouter can be used to access the Anthropic, OpenAI, and Google models listed above. Other custom models available on OpenRouter are not currently supported in Reliant.

**Getting an API Key**:

* Sign up at [openrouter.ai](https://openrouter.ai)
* Navigate to Keys section
* Create new API key

## Configuration Priority

Reliant checks for configuration in this order:

1. **Environment Variables** (highest priority)
2. **Configuration File** (`.reliant.json`)
3. **Settings UI** (stored in config file)

Environment variables override configuration file settings, allowing temporary changes without modifying files.

## OAuth in Web Mode

Claude Code and Codex use OAuth flows that require a **localhost callback** — the provider redirects your browser to `localhost` after you sign in. In the **desktop app (Electron)**, this is handled automatically.

If you're using Reliant through a **web browser** (e.g. accessing a remote Reliant server), you need to run a small helper on your local machine first:

```bash theme={null}
reliant auth serve
```

This starts a lightweight server on `localhost:19284` that receives the OAuth callback. It does not store any credentials — it only passes the authorization code back to your browser, and the Reliant server handles the rest.

**Steps for web mode:**

1. Run `reliant auth serve` in your terminal (keep it running)
2. Open **Settings → AI** in the web UI
3. Select **Claude Code** or **Codex**
4. Click the **Login** button (it appears once the helper is detected)
5. Complete sign-in in your browser
6. You can stop `reliant auth serve` after connecting

If the helper isn't running, the UI will show the command you need to run instead of the login button.

<Info>
  You can customize the port with `reliant auth serve --port 9999`.
</Info>

## Verifying Provider Authentication

After configuring a provider:

1. Open Reliant Settings
2. Navigate to the AI Providers section
3. Find your configured provider
4. Verify connection status:
   * API key providers: click **Test Connection** or **Validate**
   * Codex provider: use **Login with Codex** and confirm success state
5. Verify you see a success message

## Common Issues

### "Invalid API Key" Error

**Solutions**:

* Verify key is copied correctly with no extra spaces
* Check key hasn't expired (regenerate if needed)
* Ensure you're using the correct provider (for example, an Anthropic key for Claude models)

### "Rate Limit Exceeded"

**Solutions**:

* Wait before sending more requests
* Check your provider's rate limits for your tier
* Consider upgrading your plan
* Use a different API key or provider temporarily

### Provider Not Showing in UI

**Solutions**:

* Restart Reliant after configuration changes
* Check configuration file syntax (JSON must be valid)
* Ensure provider isn't marked as `"disabled": true`

### Connection Timeout

**Solutions**:

* Verify internet connection
* Check firewall settings
* For local models, ensure server is running
* Disable VPN temporarily to test

## Security Best Practices

### Never Commit API Keys

Add to `.gitignore`:

```text theme={null}
.reliant.json
.env
```

### Use Environment Variables for CI/CD

In CI/CD pipelines, set API keys as secret environment variables:

```yaml theme={null}
env:
  ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
```

### Rotate Keys Regularly

* Rotate API keys every 90 days
* Immediately rotate if a key may have been exposed
* Use different keys for development and production

### Limit Key Permissions

Where possible:

* Use read-only or limited-scope keys
* Create separate keys for different projects
* Monitor key usage in provider dashboards

## Rate Limits and Billing

Different provider tiers have different limits. Refer to your provider's documentation for the most up-to-date information on rate limits and pricing:

* **Anthropic**: [Rate Limits](https://docs.anthropic.com/en/api/rate-limits) | [Pricing](https://www.anthropic.com/pricing)
* **OpenAI**: [Rate Limits](https://platform.openai.com/docs/guides/rate-limits) | [Pricing](https://openai.com/pricing)
* **Google Gemini**: [Pricing and Limits](https://ai.google.dev/pricing)
* **OpenRouter**: [Models and Pricing](https://openrouter.ai/models)

## Related Topics

* [Presets](/workflows/presets) - Configure which models workflows use
* [MCP Servers](/settings/mcp-servers) - Extend capabilities with external tools
* [Troubleshooting](/help/troubleshooting) - Solve common API key issues