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

# OpenAI API-Compatible Servers

> Learn how to integrate any OpenAI-compatible API provider with your Aikeedo platform.

## Introduction

Aikeedo supports any API provider that implements the OpenAI API specification. This means you can integrate any service that follows OpenAI's API format, whether it's a cloud service, self-hosted model, or custom implementation.

<Note>
  If a service is compatible with OpenAI's API specification, it will work with Aikeedo - there are no restrictions on which providers you can use.
</Note>

## Common Providers

While you can use any OpenAI-compatible provider, here are some popular options:

* Together AI
* OpenRouter
* Groq
* DeepSeek
* Google AI (Gemini)
* Perplexity AI
* Hugging Face
* Nebius
* DeepInfra
* AI/ML API
* Azure OpenAI Service
* v0 by Vercel
* Self-hosted models (via compatible servers)
* Custom LLM implementations

## Adding a New Provider

1. Go to Settings → Integrations
2. Scroll to "Custom LLM Servers" section
3. Click "New server"

### Configuration Fields

#### API Server

* **Name**: A descriptive name for the provider
* **Server address**: The API endpoint URL
* **API Key/Authorization token**: Your authentication token for the service

<Note>
  The server address supports dynamic variables for flexible configuration.
</Note>

##### **Dynamic Variables**

You can use the following variable in your server address:

* `{model}`: Will be dynamically replaced with the current model name

**Example for Hugging Face**:

```text theme={null}
https://api-inference.huggingface.co/models/{model}
```

When using model "mistralai/Mistral-7B-Instruct-v0.1", the actual request will go to:

```text theme={null}
https://api-inference.huggingface.co/models/mistralai/Mistral-7B-Instruct-v0.1
```

<Tip>
  Dynamic variables are particularly useful for services like Hugging Face where the model name is part of the API endpoint.
</Tip>

#### Headers

Aikeedo automatically adds these default headers:

* `Content-Type: application/json`
* `Authorization: Bearer YOUR_API_KEY` (when API key field is filled)

<Note>
  You can override these default headers or add custom headers as needed. This is useful when a provider requires specific header configurations.
</Note>

##### Adding Custom Headers

1. Click "Add header"
2. Enter the header key (e.g., `HTTP-Referer`)
3. Enter the header value
4. Repeat for additional headers

##### Examples

**Override default Authorization header**:

```text theme={null}
Key: Authorization
Value: Basic YOUR_BASE64_CREDENTIALS
```

**Add provider-specific headers**:

```text theme={null}
Key: X-Title
Value: Your Application Name
```

<Tip>
  If you need to override the default Content-Type or Authorization headers, simply add them with your desired values in the custom headers section.
</Tip>

#### Models Configuration

For each model you want to use:

* **Key**: Unique identifier for the model (e.g., `gpt-3.5-turbo`)
* **Name**: Display name shown to users
* **Provider**: The model provider's name
* **Vision**: Toggle if the model supports image analysis
* **Tools**: Toggle if the model supports function calling

## Provider-Specific Setup

<AccordionGroup>
  <Accordion title="Together AI">
    ```text Server theme={null}
    https://api.together.xyz/v1
    ```

    Get your API key from [Together AI Settings](https://api.together.ai/settings/api-keys)
  </Accordion>

  <Accordion title="OpenRouter">
    ```text Server theme={null}
    https://openrouter.ai/api/v1
    ```

    Get your API key from [OpenRouter Settings](https://openrouter.ai/settings/keys)
  </Accordion>

  <Accordion title="Google AI (Gemini)">
    ```text Server theme={null}
    https://generativelanguage.googleapis.com/v1beta/openai
    ```

    Get your API key from [Google AI Studio](https://aistudio.google.com/apikey)
  </Accordion>

  <Accordion title="Hugging Face">
    ```text Server theme={null}
    https://api-inference.huggingface.co/models/{model}/v1/
    ```

    Get your API key from [Hugging Face Settings](https://huggingface.co/settings/tokens)

    <Note>
      The {model} variable in the server address will be automatically replaced with your chosen model name.
    </Note>
  </Accordion>

  <Accordion title="Groq">
    ```text Server theme={null}
    https://api.groq.com/openai/v1
    ```

    Get your API key from [Groq Console](https://console.groq.com/keys)

    <Note>
      Groq is known for its extremely fast inference speeds and competitive pricing.
    </Note>
  </Accordion>

  <Accordion title="DeepSeek">
    ```text Server theme={null}
    https://api.deepseek.com
    ```

    Required headers:

    * `Authorization`: Bearer YOUR\_API\_KEY

    Get your API key from [DeepSeek Platform](https://platform.deepseek.com/api_keys)

    <Note>
      DeepSeek offers both general-purpose and code-specialized models with competitive pricing.
    </Note>
  </Accordion>

  <Accordion title="Perplexity">
    ```text Server theme={null}
    https://api.perplexity.ai
    ```

    Get your API key from [Perplexity Settings](https://www.perplexity.ai/settings/api)

    <Note>
      Perplexity offers high-quality models with strong reasoning capabilities and up-to-date knowledge.
    </Note>
  </Accordion>

  <Accordion title="Nebius">
    ```text Server theme={null}
    https://api.studio.nebius.ai/v1/
    ```

    Get your API key from [Nebius Studio](https://studio.nebius.com/settings/api-keys)
  </Accordion>

  <Accordion title="DeepInfra">
    ```text Server theme={null}
    https://api.deepinfra.com/v1/openai/
    ```

    Get your API key from [DeepInfra Dashboard](https://deepinfra.com/dash/api_keys)

    <Note>
      DeepInfra provides access to a variety of open-source models with competitive pricing and low latency.
    </Note>
  </Accordion>

  <Accordion title="AI/ML API">
    ```text Server theme={null}
    https://api.aimlapi.com/v1/
    ```

    Get your API key from [AI/ML API Keys](https://aimlapi.com/app/keys)
  </Accordion>

  <Accordion title="Azure OpenAI Service">
    ```text Server theme={null}
    https://:resource_name.openai.azure.com/openai/deployments/:deployment_name/chat/completions?api-version=:api_version
    ```

    Get your API key and other details from [Azure Portal](https://portal.azure.com/#blade/Microsoft_Azure_ProjectOxford/CognitiveServicesHub/OpenAI)

    <Note>
      For Azure OpenAI Service:

      * Replace `:resource_name` with your Azure OpenAI resource name
      * Replace `:deployment_name` with `{model}` to use your configured model names as deployment names
      * Replace `:api_version` with your desired API version (e.g., '2024-10-21')
      * Different endpoints (like completions, embeddings) will need their specific paths
    </Note>

    <Tip>
      Example configuration:

      ```text theme={null}
      https://my-instance.openai.azure.com/openai/deployments/{model}/chat/completions?api-version=2024-10-21
      ```

      When you configure a model named "gpt-4o-mini", it will automatically use that as the deployment name in the URL.
    </Tip>
  </Accordion>

  <Accordion title="v0 by Vercel">
    ```text Server theme={null}
    https://api.v0.dev/v1
    ```

    Model name: `v0-1.0-md`

    Get your API key from [v0.dev Settings](https://v0.dev/chat/settings/keys)

    <Note>
      The v0 API is currently in beta and requires a Premium or Team plan with usage-based billing enabled.
    </Note>
  </Accordion>
</AccordionGroup>

## Tools Compatibility

When configuring models for your custom LLM server, you can enable them for:

1. **Chat**: Interactive conversations and assistance
2. **Writer**: Content generation and writing tasks
3. **Coder**: Programming help and code generation
4. **Title Generation**: Automatic title creation for content

<Note>
  Make sure to enable only the capabilities that your chosen model actually supports. Enabling unsupported features may result in unexpected behavior.
</Note>

## Best Practices

1. **Testing**:
   * Test each model after configuration
   * Verify response formats
   * Check token limits and pricing

2. **Security**:
   * Keep API keys secure
   * Use HTTPS for external providers
   * Regularly rotate API keys

3. **Monitoring**:
   * Track API usage
   * Monitor response times
   * Check for error rates

## Troubleshooting

Common issues and solutions:

1. **Authentication Errors**:
   * Verify API key format
   * Check header configuration
   * Confirm server address is correct

2. **Model Issues**:
   * Ensure model names match provider's specifications
   * Verify model availability in your subscription
   * Check provider's status page

3. **Connection Problems**:
   * Verify network connectivity
   * Check for firewall restrictions
   * Confirm server address format

<Note>
  Always refer to your provider's documentation for the most up-to-date configuration details and troubleshooting guides.
</Note>

## Rate Limits and Quotas

* Monitor your provider's rate limits
* Check quota usage regularly
* Set up alerts for quota thresholds
* Consider implementing retry logic for rate limit errors

## Need Help?

If you need assistance with Aikeedo:

<CardGroup cols={2}>
  <Card title="Professional Support" icon="headset" href="https://aikeedo.com/support/">
    Get expert help from our team with a paid support subscription
  </Card>

  <Card title="Troubleshooting Guide" icon="wrench" href="/setup/troubleshooting">
    Check common issues and solutions
  </Card>
</CardGroup>

<Note>
  Keep your integrations updated as providers may change their API specifications or requirements.
</Note>