claude-code

建议使用智谱，国产，包年仅需240元，日常轻量使用够用。

安装claude-code

npm install -g @anthropic-ai/claude-code

设置环境变量

我的电脑为macOS系统。仅供参考。
编辑 ~/.zshrc

export ANTHROPIC_BASE_URL="https://open.bigmodel.cn/api/anthropic"
export ANTHROPIC_AUTH_TOKEN="aaa.bbb"

export ANTHROPIC_MODEL="glm-4.6"
export ANTHROPIC_DEFAULT_OPUS_MODEL="glm-4.6"
export ANTHROPIC_DEFAULT_SONNET_MODEL="glm-4.6"
export ANTHROPIC_SMALL_FAST_MODEL="glm-4.5-air"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="glm-4.5-air"

export CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC="1"

编辑 ~/.claude/config.json

{
  "primaryApiKey": "default"
}

然后就可以在terminal或者powershell/cmd内执行claude了。

docs
npm
智谱

codex

建议使用 具有缓存功能的模型或中转站。

安装codex

npm install -g @openai/codex

创建配置文件 与 设置环境变量

我的电脑为macOS系统。仅供参考。
vi ~/.codex/config.toml

model_provider = "deepseek"
model = "deepseek-chat"
# model_provider = "shareyourcc"
# model = "gpt-5" 
# model_reasoning_effort = "high"

# network_access = "enabled"
disable_response_storage = true


[model_providers.shareyourcc]
name = "shareyourcc"
base_url = "https://api-us.shareyour.cc/v1"
env_key = "CODEX_API_KEY_SHAREYOURCC"    
wire_api = "responses"
# requires_openai_auth = true

[model_providers.tuzi]
name = "tuzi"
base_url = "https://api.tu-zi.com/v1"
env_key = "CODEX_API_KEY_TUZI"    
wire_api = "responses"

[model_providers.duck]
name = "duck"
base_url = "https://jp.duckcoding.com/v1"
env_key = "CODEX_API_KEY_DUCKCODING"    
wire_api = "responses"

[model_providers.deepseek]
name = "deepseek"
base_url = "https://api.deepseek.com"
env_key = "CODEX_API_KEY_DEEPSEEK"    
wire_api = "chat"

编辑 ~/.zshrc:

export CODEX_API_KEY_TUZI="sk-aaa"
export CODEX_API_KEY_SHAREYOURCC="sk-user-bbb"
export CODEX_API_KEY_DUCKCODING="sk-ccc"
export CODEX_API_KEY_DEEPSEEK="sk-ddd"

编辑 ~/.codex/auth.json

{
  "OPENAI_API_KEY": null
}

然后就可以在terminal或者powershell/cmd内执行codex了。

npm
config