Windows下如何搭建ClaudeCode调用本地Qwen3-code

先介绍一下系统环境

大模型搭建在局域网的一台ubuntu 24.04.3上

LLM的机器配置：

4块A100-40Gb的GPU卡

Ollama 0.12.6

litellm 1.78.7

客户机配置：

windows11

ollama配置

ollama.com/

ollama安装略了，网上一堆，这里只说明重要的配置主要是监听IP开放，并追加了日志已方便调试，调试没问题可以将监听地址改为局域网

bash 复制代码

# cat /etc/systemd/system/ollama.service
# 这是ollama的系统服务脚本
[Unit]
Description=Ollama Service
After=network-online.target

[Service]
ExecStart=/usr/local/bin/ollama serve
Environment="OLLAMA_HOST=0.0.0.0"
Environment="OLLAMA_ORIGINS=*"
Environment="STREAMING=true"
User=ollama
Group=ollama
Restart=always
RestartSec=3

# 标准输出和错误输出重定向到文件
StandardOutput=file:/var/log/ollama/ollama.log
StandardError=file:/var/log/ollama/ollama-error.log
# 或者使用追加模式
# StandardOutput=append:/var/log/ollama/ollama.log
# StandardError=append:/var/log/ollama/ollama-error.log

# 日志文件限制
LogRateLimitIntervalSec=30
LogRateLimitBurst=1000

Environment="PATH=/root/.nvm/versions/node/v22.20.0/bin:/root/anaconda3/bin:/root/anaconda3/condabin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/usr/bin/gcc:/root/anaconda3/bin:/usr/local/matlab/bin"

[Install]
WantedBy=default.target

litellm 配置

github.com/BerriAI/lit...

这里只说明重要的配置

bash 复制代码

# cat config.yaml
# 为什么这么配置，主要是claude code会调用自家的模型，这里都做了转换为qwen3-code
model_list:
  - model_name: "qwen3-code"
    litellm_params:
      model: "ollama/qwen3-coder:latest" 
      api_base: "http://127.0.0.1:11434"
      api_key: "none"
  # 添加 Anthropic 兼容的模型别名
  - model_name: "claude-3-5-sonnet-20241022"
    litellm_params:
      model: "ollama/qwen3-coder:latest"
      api_base: "http://127.0.0.1:11434"
      api_key: "none"
  
  - model_name: "claude-3-5-haiku-20241022"
    litellm_params:
      model: "ollama/qwen3-coder:latest"
      api_base: "http://127.0.0.1:11434"
      api_key: "none"
  
  - model_name: "claude-3-opus-20240229"
    litellm_params:
      model: "ollama/qwen3-coder:latest"
      api_base: "http://127.0.0.1:11434"
      api_key: "none"

  - model_name: "claude-haiku-4-5-20251001"
    litellm_params:
      model: "ollama/qwen3-coder:latest"
      api_base: "http://127.0.0.1:11434"
      api_key: "none"

bash 复制代码

# cat cat /etc/systemd/system/litellm.service
# 这是litellm的系统服务脚本
[Unit]
Description=LiteLLM Proxy Server
After=network.target

[Service]
Type=simple
User=pan
WorkingDirectory=/home/pan/LiteLLM
ExecStart=/home/pan/LiteLLM/start_litellm.sh
Restart=always
RestartSec=10

[Install]
WantedBy=multi-user.target

bash 复制代码

# cat /home/pan/LiteLLM/start_litellm.sh 
#!/bin/bash
# LiteLLM 启动脚本
# 这里监听本机IP并指定8000端口为服务端口
cd /home/pan/LiteLLM
source .venv/bin/activate
exec litellm --config config.yaml --port 8000 --host 192.168.84.251

好了，上面介绍了ollama+litellm的组合配置，这里ollama对外服务是开放的，只是为了测试ollama的服务是否正常，投入使用后，请将0.0.0.0改为127.0.0.1，只供litellm访问。

防火墙配置

bash 复制代码

# 检查防火墙状态
sudo ufw status

# 如果启用，确保端口开放
sudo ufw allow 11434/tcp
sudo ufw allow 8000/tcp

测试服务是否正常（在ubuntu机器上进行测试）

ollama

bash 复制代码

curl -v http://192.168.84.251:11434/api/tags

# 返回：模型列表，说明ollama服务正常

*   Trying 192.168.84.251:11434...
* Connected to 192.168.84.251 (192.168.84.251) port 11434
> GET /api/tags HTTP/1.1
> Host: 192.168.84.251:11434
> User-Agent: curl/8.9.1
> Accept: */*
> 
* Request completely sent off
< HTTP/1.1 200 OK
< Content-Type: application/json; charset=utf-8
< Date: Tue, 04 Nov 2025 10:02:07 GMT
< Transfer-Encoding: chunked
< 
{"models":[{"name":"qwen3-coder:latest","model":"qwen3-coder:latest".....

litellm

bash 复制代码

curl http://192.168.84.251:8000/v1/models

# 返回：模型列表，说明litellm服务正常
{"data":[{"id":"qwen3-code","object":"model","created":1677610602,"owned_by":"openai"}......

在windows机器上安装claude code

安装claude code 用PowerShell打开系统管理员终端：

docs.claude.com/en/docs/cla...

powershell 复制代码

# 需要node.js 18+
npm install -g @anthropic-ai/claude-code

验证安装，

powershell 复制代码

mkdir claude_test ; cd claude_test
claude

提示连接anthropic.com官网失败：

powershell 复制代码

Welcome to Claude Code v2.0.32
..............................................................................................................................................................................

     *                                       █████▓▓░
                                 *         ███▓░     ░░
            ░░░░░░                        ███▓░
    ░░░   ░░░░░░░░░░                      ███▓░
   ░░░░░░░░░░░░░░░░░░░    *                ██▓░░      ▓
                                             ░▓▓███▓▓░
 *                                 ░░░░
                                 ░░░░░░░░
                               ░░░░░░░░░░░░░░░░
       █████████                                        *
      ██▄█████▄██                        *
       █████████      *
.....................█ █   █ █..............................................................................................................................

 Unable to connect to Anthropic services

 Failed to connect to api.anthropic.com: ERR_BAD_REQUEST

 Please check your internet connection and network settings.

 Note: Claude Code might not be available in your country. Check supported countries at
 https://anthropic.com/supported-countries
PS C:\Users\MSI\claude_test>

给claude code添加本地环境：

说明：创建全局配置文件，这里指定模型为litellm的config.yaml配置的模型，前提ollama下载了该模型，同时指定了base_url的地址。

PowerShell 复制代码

echo '` {
  "env": {
    "ANTHROPIC_MODEL": "qwen3-code",
    "ANTHROPIC_SMALL_FAST_MODEL": "qwen3-code",
    "ANTHROPIC_BASE_URL": "http://192.168.84.251:8000",
    "ANTHROPIC_AUTH_TOKEN": "litellm"
  }
}' > C:\Users\MSI\.claude\settings.json

再次运行claude

PowerShell 复制代码

Welcome to Claude Code v2.0.32
..............................................................................................................................................................................

     *                                       █████▓▓░
                                 *         ███▓░     ░░
            ░░░░░░                        ███▓░
    ░░░   ░░░░░░░░░░                      ███▓░
   ░░░░░░░░░░░░░░░░░░░    *                ██▓░░      ▓
                                             ░▓▓███▓▓░
 *                                 ░░░░
                                 ░░░░░░░░
                               ░░░░░░░░░░░░░░░░
       █████████                                        *
      ██▄█████▄██                        *
       █████████      *
.....................█ █   █ █..............................................................................................................................

 Let's get started.

 Choose the text style that looks best with your terminal
 To change this later, run /theme

 ❯ 1. Dark mode ✔
   2. Light mode
   3. Dark mode (colorblind-friendly)
   4. Light mode (colorblind-friendly)
   5. Dark mode (ANSI colors only)
   6. Light mode (ANSI colors only)

╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌
  1    function greet() {
  2 -    console.log("Hello, World!");
  2 +    console.log("Hello, Claude!");
  3    }
╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌

可以使用了。