Models and limits

Models

The model list comes from the free-pi server when the CLI starts, so it can change without a new CLI release. The startup screen shows the model currently selected. Run /model in the agent to switch to another model in the list. The picker shows only free-pi models, and shows how far each model stretches your daily allowance.

All requests go through the free-pi server. You never handle a model API key. If you set up your own provider keys in your environment, they are hidden from the picker by default, so a stray key cannot route a session around free-pi. Set FREEPI_KEEP_ENV_PROVIDERS=1 to show them.

The free daily allowance

Each account gets a free allowance every day. The size of the allowance is set by the server and can change, so these docs do not state a fixed amount. Account age can also affect it.

Run /usage to see how much of today's allowance you have used. free-pi shows it as a percentage, together with your account tier and token counts. From 80% used, the usage meter in the terminal shows a heads-up.

Other limits

• One session at a time. The server allows one active CLI session per account. See Using the agent.
• Server messages. When the server rejects a request, for example when you have reached a limit, the CLI shows one readable line with the server's message. pi may also print its raw error and retry a few times. This is expected.

When you reach the limit

Requests are rejected until the allowance resets the next day. The terminal shows the server's message. Slash commands such as /usage still work, because they do not need a model turn.