Installation

Sign in and connect a model

Install fate first, then sign in to your fate account. Your model membership and fate plan have separate accounts and charges.

1. Open fate and choose Sign in or sign up. Use Google or an email verification code in your browser, approve, then return to the app. You can cancel the wait and start again.
2. Choose your ChatGPT or Claude membership, or enter a supported provider’s API Key, in the welcome card or Settings > Account & models.
3. For a membership, install its official CLI using the links below, then sign in within fate. API Keys need no CLI and are billed by the model provider for usage.

Official installation guides for memberships

These install the model provider’s official tools. Return to fate to sign in afterwards; you do not launch fate from a terminal.

CLI not found or version not recognized

1. Open Settings > Diagnostics and check the path and version under Official command line. After installing or updating the official tool, choose Detect again.
2. If it is still missing, select the installed codex or claude executable. Canceling the file picker leaves settings intact; clearing a chosen path restores automatic discovery.
3. Correct or clear an explicitly selected path that no longer works. Adding a channel preserves the selected model; switch explicitly in the composer if needed.

Model connection failed

1. Check the model account, allowance and connection. Under Settings > Diagnostics > Model network, choose System, Direct, a custom HTTP/HTTPS proxy, or import from your terminal.
2. Enter a custom proxy’s address, username and password separately. If the proxy is unavailable, use a working HTTP/HTTPS proxy or direct connection, then retry.
3. Finish or cancel an active reading or sign-in before changing model settings. On failure the previous working settings are retained; fate does not silently change models or billing sources.