一、创建deepseek API在DeepSeek开放平台注册申请自己的API keysDeepSeek二、PowerShell运行脚本接入API根据DeepSeek官方发布的API文档接入 Codex | DeepSeek API Docs运行前请确保已安装 Codex CLI 或 ChatGPT 桌面端并至少运行过一次即~/.codex目录已存在。Windows 用户在 PowerShell 中执行irm https://cdn.deepseek.com/api-docs/codex-deepseek-setup-en.ps1 | iex实际运行会报错irm 不是内部或外部命令也不是可运行的程序 或批处理文件。解决方法调用Powershell执行powershell -ExecutionPolicy Bypass -Command irm https://cdn.deepseek.com/api-docs/codex-deepseek-setup-en.ps1 | iex执行后按照菜单提示选择模型并输入你的 DeepSeek API Key以sk-开头脚本就会自动完成所有配置。三、模型切换至DeepSeek-v4-prodeepSeek-v4-pro已正式发布可以将模型从flash切换至pro。再次调用Powershell运行powershell -ExecutionPolicy Bypass -Command irm https://cdn.deepseek.com/api-docs/codex-deepseek-setup-en.ps1 | iex选择第2个即可成功切换。Codex DeepSeek Setup v1.1.0 Codex directory: C:\Users\c\.codex Choose an action: 1. Configure Codex to use the deepseek-v4-flash model 2. Configure Codex to use the deepseek-v4-pro model 3. Restore the default Codex configuration (remove deepseek settings) Enter 1 / 2 / 3: 2 Switching default model - deepseek-v4-pro Backup and models.json from this script detected; only the model field in config.toml will be updated. [OK] config.toml updated: model deepseek-v4-pro