Custom Endpoints

You can use the OpenAI provider with any OpenAI-compatible API endpoint. This is useful for connecting to local model servers, third-party providers, or self-hosted models that expose an OpenAI-compatible chat completions API.

Configuration

Set the Endpoint option on the OpenAI provider to point to your custom API:

ASP.NET

Program.cs

builder.Services.AddRevealAI()
    .AddOpenAI(options =>
    {
        options.ApiKey = "your-api-key";
        options.Endpoint = "https://your-custom-endpoint.com/v1";
        options.Model = "your-model-name";
    });

Or using appsettings.json:

appsettings.json

{
  "RevealAI": {
    "OpenAI": {
      "ApiKey": "your-api-key",
      "Endpoint": "https://your-custom-endpoint.com/v1",
      "Model": "your-model-name"
    }
  }
}

Node.js

server.js

const aiSettings = {
    openai: {
        ApiKey: 'your-api-key',
        Endpoint: 'https://your-custom-endpoint.com/v1',
        Model: 'your-model-name'
    }
};

revealAI.withOptions({
    defaultProvider: 'openai',
    settings: aiSettings
});

Java

Application.java

Map<String, Object> aiSettings = Map.of(
        "openai", Map.of(
                "ApiKey", "your-api-key",
                "Endpoint", "https://your-custom-endpoint.com/v1",
                "Model", "your-model-name"));

Common Use Cases

Local Model Servers

Tools like Ollama, LM Studio, and vLLM expose OpenAI-compatible APIs that can be used directly:

ASP.NET

// Ollama example
builder.Services.AddRevealAI()
    .AddOpenAI(options =>
    {
        options.ApiKey = "ollama";  // Ollama doesn't require a real key
        options.Endpoint = "http://localhost:11434/v1";
        options.Model = "llama3";
    });

Node.js

// Ollama example
const aiSettings = {
    openai: {
        ApiKey: 'ollama',  // Ollama doesn't require a real key
        Endpoint: 'http://localhost:11434/v1',
        Model: 'llama3'
    }
};

Java

// Ollama example
Map<String, Object> aiSettings = Map.of(
        "openai", Map.of(
                "ApiKey", "ollama",  // Ollama doesn't require a real key
                "Endpoint", "http://localhost:11434/v1",
                "Model", "llama3"));

Third-Party Providers

Many AI providers offer OpenAI-compatible endpoints:

ASP.NET

builder.Services.AddRevealAI()
    .AddOpenAI(options =>
    {
        options.ApiKey = "your-provider-api-key";
        options.Endpoint = "https://api.provider.com/v1";
        options.Model = "provider-model-name";
    });

Node.js

const aiSettings = {
    openai: {
        ApiKey: 'your-provider-api-key',
        Endpoint: 'https://api.provider.com/v1',
        Model: 'provider-model-name'
    }
};

Java

Map<String, Object> aiSettings = Map.of(
        "openai", Map.of(
                "ApiKey", "your-provider-api-key",
                "Endpoint", "https://api.provider.com/v1",
                "Model", "provider-model-name"));

Limitations

When using custom endpoints, be aware of:

• The endpoint must support the OpenAI Chat Completions API format
• Token usage reporting depends on the endpoint's compatibility
• Reasoning model detection is based on model name patterns (o1, o3, o4, gpt-5) and may not apply to custom models
• Some advanced features may not be supported by all custom endpoints

Custom Providers

If your endpoint doesn't follow the OpenAI-compatible API format, consider building a custom provider instead.