Skip to main content
How you log in depends on how you access Poolside models.
Looking for the fastest way to get a free Poolside API key? Choose Use Poolside Platform for free (recommended).

​
Before you start

The steps on this page use Poolside Agent CLI. See Install Poolside Agent CLI. If you need an API key for the Poolside API or a third-party tool, see Authenticate API requests.

​
Choose a login option

Open a terminal and run pool login. Choose how you want to access models with the Poolside agent. To use a different ACP agent, run pool -s. After you log in, the CLI stores the credentials so you do not need to enter them every time you start pool. Choose this option if you want the fastest way to get free developer access. Sign in to Poolside Platform, then copy the API key it creates for you.
  1. In the terminal, choose Use Poolside Platform for free (recommended).
  2. In the browser window that opens, sign in to Poolside Platform or create an account.
  3. Copy the API key that Poolside creates for you.
  4. Return to the terminal.
  5. Paste the key at the Enter API key prompt.
The API key created through Poolside Platform is for Poolside-hosted inference. If your organization has a Poolside deployment, use that URL and sign-in method instead.

​
Use your organization’s Poolside deployment (enterprise)

Choose this option if your organization has its own Poolside deployment with Poolside Console. Your Poolside administrator gives you the API URL.
  1. In the terminal, choose Use your organization’s Poolside deployment (enterprise).
  2. Enter the API URL provided by your organization. Do not enter the deployment’s /openai/v1 API base URL.
  3. Choose how to authenticate:
    • Log in via browser: Complete the sign-in in the browser window that opens.
    • Provide API token: Paste your token at the Enter API token prompt.
For more information, see the Poolside Console documentation.

​
Use your OpenRouter account

Choose this option if you already use OpenRouter or you need paid access to Poolside models with a larger context window. To see which Poolside models OpenRouter offers, go to Poolside models on OpenRouter.
  1. In the terminal, choose Use your OpenRouter account.
  2. Go to OpenRouter API keys, sign in, and create an API key.
  3. Return to the terminal.
  4. Paste the key at the Enter API key prompt.

​
Connect an OpenAI-compatible provider

Choose this option for self-managed Poolside inference, an inference gateway such as Bifrost or LiteLLM, a third-party provider, or a model server such as vLLM or Ollama.
  1. In the terminal, choose Connect an OpenAI-compatible provider.
  2. Enter the API base URL supplied by your provider or administrator. A direct self-managed Poolside model endpoint uses https://<model-hostname>/v1.
  3. Paste the endpoint’s API key at the Enter API key prompt.
See API base URLs by access method.
If the endpoint does not validate API keys, enter any non-empty value, such as local-test.If the endpoint does not list its models, specify one when starting pool:

​
Verify your login

Run:
When your credentials work, pool starts an interactive session and shows a Connected to agent server: message. If the CLI cannot connect, see Troubleshoot Poolside Agent CLI.

​
Switch accounts or providers

Log out, then log in again with a different option:

​
Next steps