Qwen3开源版 extra_body 传递失败

[yuanhongyi]

官方示例

import os
from openai import OpenAI


client = OpenAI(
    # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx",
    api_key="123",
    base_url="http://localhost:9090/v1",
)

completion = client.chat.completions.create(
    # 模型列表：https://help.aliyun.com/zh/model-studio/getting-started/models
    model="qwen-plus",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "你是谁？"},
    ],
    # Qwen3模型通过enable_thinking参数控制思考过程（开源版默认True，商业版默认False）
    # 使用Qwen3开源版模型时，若未启用流式输出，请将下行取消注释，否则会报错
    extra_body={"enable_thinking": False},
)
print(completion.model_dump_json())

按照如上例子给自建服务发请求，得到如下报错

openai.InternalServerError: Error code: 500 - {'error': 'Post "https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions": HTTP error: 400 Bad Request, body: {"error":{"code":"invalid_parameter_error","param":null,"message":"parameter.enable_thinking must be set to false for non-streaming calls","type":"invalid_request_error"},"id":"chatcmpl-6d65201b-cad4-93bd-a49e-181cc76aa757","request_id":"6d65201b-cad4-93bd-a49e-181cc76aa757"}'}

但是改为官方base_url，就不会有这样的报错

这是为什么？