#gpt-4

Integrate GPT-4 API effectively. Patterns: 1. Chat completions with system/user messages. 2. Function calling for structured outputs. 3. Streaming responses for better UX. 4. Token counting to manage costs. 5. Temperature and top_p tuning. 6. Max tokens control. 7. Error handling and retries. 8. Rate limiting awareness. Use tiktoken for accurate token counts and implement caching for repeated queries.