Connect CodeX to 211API

Add a provider pointing at 211API in the CodeX config file and the gpt-5.6 model family is ready — your task orchestration habits stay unchanged.

Prerequisites

• A 211API account with an API Key created in the console (starts with sk-).
• Node.js 18 or later installed locally.

Step 1 · Install CodeX

npm install -g @openai/codex

Step 2 · Configure the provider

Edit ~/.codex/config.toml (on Windows: %USERPROFILE%\.codex\config.toml) and register 211API as a model provider:

model = "gpt-5.6-sol"
model_provider = "211api"

[model_providers.211api]
name = "211API"
base_url = "https://www.211api.com/v1"
env_key = "211API_API_KEY"

Then put the API Key into an environment variable. macOS / Linux:

export 211API_API_KEY="sk-your-211API-key"

Windows (PowerShell):

$env:211API_API_KEY = "sk-your-211API-key"

# Persist at user level:
# setx 211API_API_KEY "sk-your-211API-key"

Note

gpt-5.6-sol targets complex reasoning and long tasks. For high-frequency light calls, switch model in config.toml to gpt-5.6-luna, which has a lower channel multiplier. Exact model names follow the console "Models" list.

Step 3 · Launch & verify

Send a simple request from your project directory to confirm auth and model routing:

cd your-project
codex "Describe this project's structure in one sentence"

A normal response means you're connected. Call logs, token usage and multiplier details are visible entry by entry in the 211API console.

FAQ

Model not found — The model value must exactly match the 211API console model list, including case and suffixes.

Environment variable not applied — On Windows, setx requires a new terminal. Inside PowerShell, read it back with $env:211API_API_KEY to confirm.