Langroid Files

OpenAI HTTP Client Configuration Support

This release introduces comprehensive HTTP client configuration support for OpenAI models, enabling usage in corporate environments with SSL certificate verification issues and custom proxy requirements.

Key Changes:

• Three-tier configuration system: Simple SSL bypass, configuration dictionary, and custom factory options
• SSL verification control: New http_verify_ssl parameter for quick SSL bypass
• Advanced configuration: http_client_config for proxy, timeout, and SSL settings
• Custom client factory: http_client_factory for complete control over HTTP client creation
• Intelligent caching: Configuration-based clients are cached for performance

Problem Solved:

Users in corporate environments with self-signed certificates or proxy requirements could not use OpenAI models due to SSL verification failures.

Usage Examples:

:::python
# Simple SSL bypass (cacheable)
config = OpenAIGPTConfig(http_verify_ssl=False)

# Configuration dictionary (cacheable)
config = OpenAIGPTConfig(http_client_config={"verify": False, "proxy": "http://proxy:8080"})

# Custom factory (not cacheable)
config = OpenAIGPTConfig(http_client_factory=lambda: httpx.Client(verify=False))

For detailed technical documentation, see: - Developer Documentation - Implementation Details

Migration:

All changes are backward compatible. No existing code needs modification.