LLM Providers

// ChatOpenAI configuration
{
  modelName: "gpt-4o-mini",
  temperature: 0.9,
  maxTokens: 2000,
  streaming: true,
  topP: 1.0,
  frequencyPenalty: 0,
  presencePenalty: 0
}

// ChatAnthropic configuration
{
  modelName: "claude-3-haiku",
  temperature: 0.9,
  maxTokensToSample: 2000,
  streaming: true,
  extendedThinking: false,
  allowImageUploads: true
}

// Enable reasoning for Claude 3.5 Sonnet or Claude 4
{
  modelName: "claude-3-5-sonnet-20241022",
  extendedThinking: true,
  budgetTokens: 1024 // Max tokens for internal reasoning
}

// ChatOllama configuration
{
  baseUrl: "http://localhost:11434",
  modelName: "llama3",
  temperature: 0.9,
  streaming: true,
  keepAlive: "5m",
  allowImageUploads: false
}

// AzureChatOpenAI configuration
{
  modelName: "gpt-4",
  temperature: 0.9,
  azureOpenAIApiDeploymentName: "my-gpt4-deployment",
  azureOpenAIApiVersion: "2024-02-15-preview"
}

// Groq configuration  
{
  modelName: "llama-3.1-70b-versatile",
  temperature: 0.9,
  maxTokens: 2000,
  streaming: true
}

// ChatGoogleGenerativeAI configuration
{
  modelName: "gemini-pro",
  temperature: 0.9,
  maxOutputTokens: 2048,
  topP: 1.0,
  topK: 40
}

// Enable image uploads
{
  allowImageUploads: true,
  imageResolution: "high" // OpenAI: low, high, auto
}

// OpenAI o1/o3 configuration
{
  modelName: "o1-preview",
  reasoning: true,
  reasoningEffort: "medium", // low, medium, high
  reasoningSummary: "auto" // auto, concise, detailed
}

{
  streaming: true // Works with most providers
}

// Ollama JSON mode
{
  jsonMode: true
}

// Add to system prompt:
// "Format all responses as JSON object"

// OpenAI-compatible endpoints
{
  basePath: "https://api.custom-endpoint.com/v1",
  proxyUrl: "http://proxy.company.com:8080"
}

// Add retry logic
{
  timeout: 60000, // 60 seconds
  maxRetries: 3
}

// Test Ollama connection
curl http://localhost:11434/api/tags

// Test OpenAI credentials
curl https://api.openai.com/v1/models \
  -H "Authorization: Bearer $OPENAI_API_KEY"