前言

最近做项目，比如写一些大模型的周边支持，需要类似问题解答，上下文联系支持，MCP，function calling之类的，先从简单的示例开始，比如python，其实大模型相关的文章，绝大部分是python写的，主流的框架有longchain springai等。其实笔者是为了测试自由线程的情况，所以就用了自由线程模式，结果踩坑了。

准备

ollama安装，拉取镜像，这个笔者以前写过：M4 Mac mini运行ds

本次为了省内存+提高效率使用了gemma3:4b模型

国内可以使用ModelScope社区加速

demo

执行pip install ollama，当然ollama兼容OpenAI，也可以使用openai的包。效果一样

import asyncio
from ollama import AsyncClient

async def chat():
    message = {'role': 'user', 'content': '你是谁?'}
    async for part in await AsyncClient().chat(model='gemma3:4b', messages=[message], stream=True):
        print(part['message']['content'], end='', flush=True)



if __name__ == '__main__':
    asyncio.run(chat())

用个异步流式方式打个招呼，代码本身没啥问题，ide也没有报错

/usr/local/bin/python /Users/huahua/PycharmProjects/chat-demo/chat.py 
我是 Gemma，一个开放权重的 AI 助手。我是一个由 Google DeepMind 训练的大型语言模型。

我是一个公开可用的模型，这意味着任何人都可以访问和使用我。

我擅长理解和生成文本，可以进行对话、回答问题、提供摘要，甚至可以尝试进行创意写作。

请注意，我没有访问外部工具或实时信息的能力。
Process finished with exit code 0

但是笔者注意到，用的非自由线程的版本，于是出事了。

问题

当笔者手工执行命令时：

使用自由线程命令居然不能读取到pip安装的ollama包。后面问AI，说是需要指定解释器

python -m pip install ollama

huahua@huahuadeMacBook-Air chat-demo % python -m pip install ollama
Collecting ollama
  Using cached ollama-0.6.1-py3-none-any.whl.metadata (4.3 kB)
Collecting httpx>=0.27 (from ollama)
  Using cached httpx-0.28.1-py3-none-any.whl.metadata (7.1 kB)
Collecting pydantic>=2.9 (from ollama)
  Using cached pydantic-2.12.5-py3-none-any.whl.metadata (90 kB)
Collecting anyio (from httpx>=0.27->ollama)
  Using cached anyio-4.12.1-py3-none-any.whl.metadata (4.3 kB)
Collecting certifi (from httpx>=0.27->ollama)
  Using cached certifi-2026.1.4-py3-none-any.whl.metadata (2.5 kB)
Collecting httpcore==1.* (from httpx>=0.27->ollama)
  Using cached httpcore-1.0.9-py3-none-any.whl.metadata (21 kB)
Collecting idna (from httpx>=0.27->ollama)
  Using cached idna-3.11-py3-none-any.whl.metadata (8.4 kB)
Collecting h11>=0.16 (from httpcore==1.*->httpx>=0.27->ollama)
  Using cached h11-0.16.0-py3-none-any.whl.metadata (8.3 kB)
Collecting annotated-types>=0.6.0 (from pydantic>=2.9->ollama)
  Using cached annotated_types-0.7.0-py3-none-any.whl.metadata (15 kB)
Collecting pydantic-core==2.41.5 (from pydantic>=2.9->ollama)
  Using cached pydantic_core-2.41.5-cp314-cp314t-macosx_11_0_arm64.whl.metadata (7.3 kB)
Collecting typing-extensions>=4.14.1 (from pydantic>=2.9->ollama)
  Using cached typing_extensions-4.15.0-py3-none-any.whl.metadata (3.3 kB)
Collecting typing-inspection>=0.4.2 (from pydantic>=2.9->ollama)
  Using cached typing_inspection-0.4.2-py3-none-any.whl.metadata (2.6 kB)
Using cached ollama-0.6.1-py3-none-any.whl (14 kB)
Using cached httpx-0.28.1-py3-none-any.whl (73 kB)
Using cached httpcore-1.0.9-py3-none-any.whl (78 kB)
Using cached h11-0.16.0-py3-none-any.whl (37 kB)
Using cached pydantic-2.12.5-py3-none-any.whl (463 kB)
Using cached pydantic_core-2.41.5-cp314-cp314t-macosx_11_0_arm64.whl (1.9 MB)
Using cached annotated_types-0.7.0-py3-none-any.whl (13 kB)
Using cached typing_extensions-4.15.0-py3-none-any.whl (44 kB)
Using cached typing_inspection-0.4.2-py3-none-any.whl (14 kB)
Using cached anyio-4.12.1-py3-none-any.whl (113 kB)
Using cached idna-3.11-py3-none-any.whl (71 kB)
Using cached certifi-2026.1.4-py3-none-any.whl (152 kB)
Installing collected packages: typing-extensions, idna, h11, certifi, annotated-types, typing-inspection, pydantic-core, httpcore, anyio, pydantic, httpx, ollama
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╺━━━━━━━━━  9/12 [pydantic]  WARNING: The script httpx is installed in '/Library/Frameworks/PythonT.framework/Versions/3.14/bin' which is not on PATH.
  Consider adding this directory to PATH or, if you prefer to suppress this warning, use --no-warn-script-location.
Successfully installed annotated-types-0.7.0 anyio-4.12.1 certifi-2026.1.4 h11-0.16.0 httpcore-1.0.9 httpx-0.28.1 idna-3.11 ollama-0.6.1 pydantic-2.12.5 pydantic-core-2.41.5 typing-extensions-4.15.0 typing-inspection-0.4.2

[notice] A new release of pip is available: 25.3 -> 26.0.1
[notice] To update, run: python3t -m pip install --upgrade pip
huahua@huahuadeMacBook-Air chat-demo % python3t -m pip install --upgrade pip
Requirement already satisfied: pip in /Library/Frameworks/PythonT.framework/Versions/3.14/lib/python3.14t/site-packages (25.3)
Collecting pip
  Using cached pip-26.0.1-py3-none-any.whl.metadata (4.7 kB)
Using cached pip-26.0.1-py3-none-any.whl (1.8 MB)
Installing collected packages: pip
  Attempting uninstall: pip
    Found existing installation: pip 25.3
    Uninstalling pip-25.3:
      Successfully uninstalled pip-25.3
  WARNING: The scripts pip, pip3 and pip3.14 are installed in '/Library/Frameworks/PythonT.framework/Versions/3.14/bin' which is not on PATH.
  Consider adding this directory to PATH or, if you prefer to suppress this warning, use --no-warn-script-location.
Successfully installed pip-26.0.1

直到看到这里

恍然大悟，笔者虽然只有一个python 3.14，但是自由线程也算是不同的版本，他们是2个指令，默认情况只能使用path路径下python的文件对应的模块引入依赖。坑啊，同一个包，自由线程是"不同的"版本。

笔者已经把python默认命令指向了自由线程版本

总结

笔者之前实际上是用spring ai方式写的大模型周边逻辑，用来体验聊天 mcp服务之类的，但是spring ai项目有点啰嗦，搭建比较繁琐，就想着python不是自由线程了，结果立马跳坑，自由线程是“不同的”版本，所以需要：使用解释器安装包