ChatGPT & Claude API Complete Tutorial

Deep learning of OpenAI GPT-4o and Anthropic Claude 3.5 API calling methods, master conversation generation, streaming output, function calling and other advanced features

Supported Models

GPT-4o

gpt-4o
128K tokens

The most powerful multimodal model, supporting text and images

$5.00 / 1M tokens
Vision UnderstandingFunction CallingJSON ModeStreaming Output

GPT-4o-mini

gpt-4o-mini
128K tokens

The most cost-effective intelligent model

$0.15 / 1M tokens
Fast ResponseCost OptimizationFunction CallingStreaming Output

GPT-3.5 Turbo

gpt-3.5-turbo
16K tokens

Classic and efficient conversation model

$0.50 / 1M tokens
Fast GenerationStable and ReliableStreaming Output

Claude 3.5 Sonnet

claude-3-5-sonnet
200K tokens

Anthropic's latest flagship model with excellent programming capabilities

$3.00 / 1M tokens
Ultra-long ContextCode GenerationVision UnderstandingSafety Alignment

Claude 3.5 Haiku

claude-3-5-haiku
200K tokens

Lightweight and fast Claude model

$0.25 / 1M tokens
Fast ResponseCost OptimizationLong Text Processing

Code Examples

Basic Conversation

The simplest ChatGPT API call example

import openai

# Set API key and endpoint
openai.api_key = "YOUR_API_KEY"
openai.api_base = "https://api.example.com/v1"

# Send conversation request
response = openai.ChatCompletion.create(
    model="gpt-4o",  # Options: gpt-4o, gpt-4o-mini, gpt-3.5-turbo
    messages=[
        {"role": "system", "content": "You are a helpful AI assistant"},
        {"role": "user", "content": "Please explain what is machine learning"}
    ],
    temperature=0.7,  # Control creativity, between 0-2
    max_tokens=1000   # Maximum output length
)

print(response.choices[0].message.content)

Best Practices

Optimize Prompt Design

  • Use clear and specific instructions
  • Provide examples and format specifications
  • Set appropriate system prompts
  • Use few-shot learning techniques

Token usingOptimize

  • Set max_tokens parameter appropriately
  • Clean up conversation history in time
  • Use summaries to compress long texts
  • Choose appropriate model size

Error Handling

  • Implement retry mechanism
  • Handle Rate Limit errors
  • Validate response format
  • Set timeout duration

Security Considerations

  • Do not expose API keys on the client side
  • Implement content filtering mechanisms
  • Keep audit logs
  • Set usage limits

API Response Format

Standard response format returned by ChatGPT and Claude APIs:

{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "created": 1677858242,
  "model": "gpt-4o",
  "usage": {
    "prompt_tokens": 13,
    "completion_tokens": 17,
    "total_tokens": 30
  },
  "choices": [{
    "message": {
      "role": "assistant",
      "content": "This is the AI response content"
    },
    "finish_reason": "stop",
    "index": 0
  }]
}