Get Started with OpenCode

​

Prerequisites

​

Configure OpenCode

1

Install OpenCode in your terminal

Run the following:

curl -fsSL https://opencode.ai/install | bash

After installation, reload your shell configuration so the opencode command is available in your current session:

source ~/.zshrc

If you use bash, run source ~/.bashrc instead.

2

Select Cerebras

Run the following command, then select Cerebras from the dropdown menu:

opencode auth login

3

Configure API key

Enter your Cerebras API key:

4

Select a model and start coding!

Launch OpenCode in your project directory:

opencode

Once inside the OpenCode interface, type /models and press Enter to open the model selector. Choose a Cerebras model from the list.If a newly released model does not appear, refresh OpenCode’s model catalog, then reopen the model selector:

opencode models cerebras --refresh

Now you can start coding!

​

Available Models

opencode models --refresh

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "cerebras": {
      "models": {
        "zai-glm-4.7": {},
        "gpt-oss-120b": {}
      }
    }
  }
}

​

Troubleshooting

• 401 Unauthorized — Your API key is missing or invalid. Re-run opencode auth login, select Cerebras, and paste a fresh key from the Cloud Console.
• 403 Forbidden or “model not available” — Cerebras models in OpenCode require a paid Cerebras plan. Upgrade in the Cloud Console, then restart OpenCode.
• 404 Not Found or “model not found” — The model ID isn’t recognized by your OpenCode installation. Run opencode models --refresh to update the cached model list, then re-run /models.
• 429 Too Many Requests — You’ve hit a rate limit. Wait and retry, or check your current usage in the Cloud Console.
• Missing opencode command after install — Reload your shell (source ~/.zshrc or source ~/.bashrc) so the new binary is on your PATH.