Creates a model response for the given chat conversation using the same interface as the OpenAI Chat Completions API. Supports both streaming (when stream: true) and non-streaming modes. Request and response formats follow the OpenAI Create chat completion API. Prefer the Responses API for new integrations; this endpoint is maintained for backward compatibility.