Step-by-step setup

Set up a coding CLI and verify your key, endpoint and model step by step.

Follow this path to set up a coding CLI, connect your own key, and verify the model you selected.

OpenAI SDK and Codex base URLhttps://zashmodel.com/v1
Claude Code base URLhttps://zashmodel.com

Claude Code adds /v1/messages automatically. When sending a raw HTTP request, include /v1 exactly once in the full URL.

1. Check your current installation

Terminal
codex --version

If the CLI is already installed, you can keep it and use the manual configuration guide instead of running the installer again. Manual setup →

2. Run the site installer

The installer sets up the CLI and writes user-level gateway settings. Back up custom configuration first, and keep any backup path printed by the installer.

Terminal
curl -fsSL https://zashmodel.com/install/codex.sh | ZASHMODEL_BASE_URL="https://zashmodel.com/v1" sh

3. Paste your Zashmodel API key

When prompted, paste a key from your own Zashmodel account. Your website password, an upstream provider key, and another site’s key are not interchangeable.

API Keys →

4. Choose an accessible model

Copy the exact model ID and check its supported protocol in Model Square. An installer’s default model is not proof that your account can use it.

Browse Model Square →

Launch the CLI
codex --model YOUR_MODEL_ID

5. Verify the provider and request log

Inside the CLI, run /status and confirm the provider, endpoint and model. Send a short request, then check the serving route, status and charge in request logs.

Request logs →

Models and prices follow your account

An empty model list means there is no model available to that key yet. Check permissions and enabled routes before retrying; installation alone does not grant model access.