Codex & ChatGPT 限流突破全解：429 Too Many Requests、Usage Cap 额度用尽与排队优化

import time
import random
from openai import OpenAI, RateLimitError

client = OpenAI()

def call_codex_with_retry(prompt, max_retries=5, base_delay=1.0):
    """
    指数退避 + 全抖动 (Full Jitter) 算法：
    防止并发请求同时重试引发第二次限流风暴
    """
    for attempt in range(max_retries):
        try:
            response = client.chat.completions.create(
                model="o1-mini",
                messages=[{"role": "user", "content": prompt}]
            )
            return response.choices[0].message.content
        except RateLimitError as e:
            if attempt == max_retries - 1:
                raise e  # 超过最大重试次数抛出异常
            
            # 指数级等待时间并加入随机抖动
            sleep_time = min(32, base_delay * (2 ** attempt)) + random.uniform(0, 1)
            print(f"遇到 429 限流，等待 {sleep_time:.2f} 秒后进行第 {attempt + 1} 次重试...")
            time.sleep(sleep_time)

# 示例调用
result = call_codex_with_retry("请编写高效快速排序实现")
print(result)