AI API 超时：原因、重试模式与回退设计

# 问题：大输入 + 高 max_tokens = 响应慢
response = client.chat.completions.create(
    model="gpt-4o",
    messages=very_long_messages,  # 100K+ token
    max_tokens=4096  # 请求长输出
)

# 修复：减少输入或限制输出
response = client.chat.completions.create(
    model="gpt-4o",
    messages=trimmed_messages,  # 减少到 50K token
    max_tokens=1024  # 更短的输出
)

# 同步 — 客户端可能在等待完整响应时超时
response = client.chat.completions.create(
    model="gpt-4o",
    messages=messages,
    max_tokens=2048
)

# 流式 — 第一个 token 更快到达，保持连接活跃
stream = client.chat.completions.create(
    model="gpt-4o",
    messages=messages,
    max_tokens=2048,
    stream=True
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

import asyncio
import httpx

async def generate_video_async(client, prompt: str, timeout: int = 600):
    """提交视频生成任务并轮询完成状态。"""

    # 第 1 步：提交任务
    submit_response = await client.post(
        "/v1/video/generations",
        json={"model": "veo-3-fast", "prompt": prompt}
    )
    job_id = submit_response.json()["id"]

    # 第 2 步：轮询完成状态
    for _ in range(timeout // 5):  # 每 5 秒检查一次
        status_response = await client.get(f"/v1/video/generations/{job_id}")
        status = status_response.json()

        if status["status"] == "completed":
            return status["result"]
        elif status["status"] == "failed":
            raise RuntimeError(f"生成失败: {status.get('error')}")

        await asyncio.sleep(5)

    raise TimeoutError(f"视频生成未在 {timeout} 秒内完成")

async def call_with_fallback(messages, client, primary_model, fallback_model):
    """尝试主模型；超时后回退到更快的模型。"""
    try:
        return await asyncio.wait_for(
            client.chat.completions.create(
                model=primary_model,
                messages=messages
            ),
            timeout=30.0
        )
    except asyncio.TimeoutError:
        # 回退到更快、可能更小的模型
        return await client.chat.completions.create(
            model=fallback_model,
            messages=messages
        )

from openai import OpenAI

client = OpenAI(
    api_key="your-evolink-key",
    base_url="https://api.evolink.ai/v1"
)

# Smart Router 处理提供商选择和回退
response = client.chat.completions.create(
    model="evolink/auto",
    messages=messages
)