Set up Codex (CLI / VS Code extension)

OpenAI's coding agent. Add one custom provider to config.toml and you're set

• API format: OpenAI Responses
• Base URL format: Up to /v1 (https://gpt.cherrychat.org/v1)
• Recommended model: gpt-6.1-sol, gpt-5.5
• Platforms: macOS, Windows, Linux, VS Code

Steps

1. Install Codex (skip if installed)

   Install the CLI with a command below and check codex --version. In VS Code, search the Extensions view for "Codex" and install the one from OpenAI; it shares the same config as the CLI.

   macOS / Linux

   curl -fsSL https://chatgpt.com/codex/install.sh | sh

   Windows (PowerShell)

   powershell -ExecutionPolicy ByPass -c "irm https://chatgpt.com/codex/install.ps1 | iex"

   Or install with npm

   npm install -g @openai/codex

2. Store the key in an environment variable

   Codex reads the key from the CHERRYCHAT_API_KEY environment variable, so the config file never contains it. Reopen the terminal afterwards, and fully quit and reopen VS Code.

   • CHERRYCHAT_API_KEY (API key): sk-your-key

   macOS / Linux (append to ~/.zshrc)

   echo 'export CHERRYCHAT_API_KEY="sk-your-key"' >> ~/.zshrc
   source ~/.zshrc

   Windows (reopen PowerShell afterwards)

   setx CHERRYCHAT_API_KEY "sk-your-key"

3. Edit config.toml

   Open ~/.codex/config.toml (%USERPROFILE%\.codex\config.toml on Windows) with the command below and paste the config. The first three lines must sit at the very top, before any [section]. If the file already has a model = or model_provider = line, delete the old one; keys cannot repeat.

   • base_url (API address): https://gpt.cherrychat.org/v1 (Stop at /v1; Codex adds /responses itself)
   • model (model name): gpt-6.1-sol

   Open the file (macOS)

   mkdir -p ~/.codex && touch ~/.codex/config.toml && open -e ~/.codex/config.toml

   Open the file (Windows)

   mkdir $HOME\.codex -Force; notepad $HOME\.codex\config.toml

   config.toml content

   model = "gpt-6.1-sol"
   model_provider = "cherrychat"
   web_search = "disabled"
   
   [model_providers.cherrychat]
   name = "CherryChat"
   base_url = "https://gpt.cherrychat.org/v1"
   env_key = "CHERRYCHAT_API_KEY"
   wire_api = "responses"
   requires_openai_auth = false

4. Verify the connection

   Open a new terminal and run the command below; a reply means it works. codex exec normally runs only inside a Git repository, and --skip-git-repo-check lets this test run in any folder. You can also start codex and type /status to see the active model and provider (should be cherrychat).

5. Use it in the VS Code extension

   The extension reads the same config.toml: click the gear in the Codex sidebar → "Codex Settings" → "Open config.toml" to check it. After changing the config or the variable, fully quit and reopen VS Code.

Common pitfalls

• Set base_url to /v1 (https://gpt.cherrychat.org/v1); Codex appends /responses. Do not use https://gpt.cherrychat.org/v1/responses or the bare domain.
• wire_api must be "responses"; the "chat" value from older guides is no longer supported.
• The model, model_provider and web_search lines must come before the first [model_providers…] section, or they are read as part of it and ignored.
• Use ~/.codex/config.toml in your home folder; a project .codex/config.toml ignores model_provider. Do not name the provider openai, ollama or lmstudio (reserved).
• Keep web_search = "disabled": Codex sends its built-in web search by default, and CherryChat API does not support built-in tools yet.
• Reopen the terminal after setting the variable. On Windows, close every PowerShell and VS Code window after setx.

Error → cause → fix

Guide checked 2026-11-02. The tool's screens may change; the official docs win if they differ.

API · API Pricing

Operated by UK Company CHERRYCHAT LTD · Company Number 16096119 · Support WeChat: JsnonoChat · Content is AI-generated — please verify before relying on it