@pga-ai/adapters-llm-openai

OpenAI adapter for PGA (Genomic Self-Evolving Prompts).

Supported Models

• GPT-4 Turbo (gpt-4-turbo-preview)
• GPT-4 (gpt-4)
• GPT-3.5 Turbo (gpt-3.5-turbo)

Features

• ✅ Full LLMAdapter interface implementation
• ✅ Streaming responses support
• ✅ Token usage tracking
• ✅ Configurable temperature and top_p
• ✅ Support for custom base URLs (proxies, Azure)
• ✅ Organization ID support

Installation

npm install @pga-ai/adapters-llm-openai

Usage

Basic Usage

import { OpenAIAdapter } from '@pga-ai/adapters-llm-openai';

const adapter = new OpenAIAdapter({
  apiKey: process.env.OPENAI_API_KEY!,
  model: 'gpt-4-turbo-preview',
});

// Chat
const response = await adapter.chat([
  { role: 'user', content: 'Hello, GPT-4!' }
]);

console.log(response.content);

Streaming Responses

const adapter = new OpenAIAdapter({
  apiKey: process.env.OPENAI_API_KEY!,
  model: 'gpt-4',
});

for await (const chunk of adapter.chatStream([
  { role: 'user', content: 'Tell me a story' }
])) {
  process.stdout.write(chunk.content);
}

With PGA Genome

import { PGA } from '@pga-ai/core';
import { OpenAIAdapter } from '@pga-ai/adapters-llm-openai';
import { PostgresAdapter } from '@pga-ai/adapters-storage-postgres';

const pga = new PGA({
  llmAdapter: new OpenAIAdapter({
    apiKey: process.env.OPENAI_API_KEY!,
    model: 'gpt-4-turbo-preview',
    temperature: 0.7,
  }),
  storageAdapter: new PostgresAdapter({
    connectionString: process.env.DATABASE_URL!,
  }),
});

const genome = await pga.createGenome({
  layer0: {
    systemPrompt: 'You are a helpful AI assistant.',
    constraints: ['Be concise', 'Use examples'],
    capabilities: ['coding', 'debugging'],
  },
});

const response = await genome.chat('Help me debug this code', {
  userId: 'user-123',
});

Configuration

interface OpenAIAdapterConfig {
  /** OpenAI API Key (required) */
  apiKey: string;

  /** Model to use (default: 'gpt-4-turbo-preview') */
  model?: string;

  /** Organization ID (optional) */
  organization?: string;

  /** Base URL for proxies or Azure (optional) */
  baseURL?: string;

  /** Temperature 0-2 (default: 1.0) */
  temperature?: number;

  /** Top P 0-1 (default: 1.0) */
  topP?: number;
}

Available Models

| Model | ID | Context | Best For | |-------|-----|---------|----------| | GPT-4 Turbo | gpt-4-turbo-preview | 128K | Complex tasks, latest features | | GPT-4 | gpt-4 | 8K | High-quality responses | | GPT-3.5 Turbo | gpt-3.5-turbo | 16K | Fast, cost-effective |

Environment Variables

# Required
OPENAI_API_KEY=sk-...

# Optional
OPENAI_ORGANIZATION=org-...
OPENAI_BASE_URL=https://api.openai.com/v1

Error Handling

try {
  const response = await adapter.chat(messages);
} catch (error) {
  if (error.status === 401) {
    console.error('Invalid API key');
  } else if (error.status === 429) {
    console.error('Rate limit exceeded');
  } else if (error.status === 500) {
    console.error('OpenAI server error');
  }
}

Token Usage

const response = await adapter.chat(messages);

console.log('Usage:', {
  prompt: response.usage.promptTokens,
  completion: response.usage.completionTokens,
  total: response.usage.totalTokens,
});

Advanced Configuration

Custom Base URL (Azure OpenAI)

const adapter = new OpenAIAdapter({
  apiKey: process.env.AZURE_OPENAI_KEY!,
  baseURL: 'https://your-resource.openai.azure.com',
  model: 'gpt-4',
});

Per-Request Options

const response = await adapter.chat(messages, {
  temperature: 0.5,
  topP: 0.9,
  maxTokens: 2000,
});

Comparison with Claude

| Feature | OpenAI (GPT-4) | Claude (Opus/Sonnet) | |---------|----------------|----------------------| | Context Window | 128K (Turbo) | 200K | | Streaming | ✅ | ✅ | | Function Calling | ✅ | ✅ (Tool Use) | | Vision | ✅ | ✅ | | Speed | Fast | Very Fast (Sonnet) | | Cost | Medium | Varies |

License

MIT

Author

Luis Alfredo Velasquez Duran (Germany, 2025)

Links

• PGA Core Documentation
• OpenAI API Documentation
• GitHub Repository