Main parts of the document:

1. Quickstart
   - Install SDK
   - First API call
   - Introduction to basic concepts

2. Guides
   - Text generation
   - Vision (image understanding)
   - Function calling
   - Assistants (AI assistants)
   - Fine-tuning (model fine-tuning)

3. API Reference
   - Complete parameter descriptions for all endpoints
   - Request and response formats

4. Cookbook (code example)
   -Code examples of actual application scenarios
   - cookbook.openai.com

OpenAI Playground (platform.openai.com/playground):




Function:


1. Test API calls online without writing code


2. Adjust parameters (temperature, max_tokens, etc.)


3. View the code of the API request (Python/Node.js/curl)


4. Test the effect of System Prompt




Usage scenarios:


- Test Prompt effect


- Understand the impact of parameters on output


- Rapid prototyping

Most commonly used API endpoints:

POST https://api.openai.com/v1/chat/completions

Key parameters:
- model: model name (gpt-4o, gpt-4o-mini, etc.)
- messages: conversation history
- temperature: creativity (0-2, default 1)
- max_tokens: maximum output length
- stream: whether to stream output
- response_format: output format (JSON mode)

Python example:
```python
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello"}],
    temperature=0.7,
    max_tokens=1000
)
```