Choose your AI connection.

Bring your own provider

Select an OpenAI-compatible endpoint, including LiteLLM, or Gemini in extension settings. Supply the provider’s base URL, key and supported model name. BYOK works without signing in to xgoose.

Inference requests go to the connection you configure. Your provider’s pricing, access rules and privacy terms apply. Uploading BYOK conversations to xgoose is optional, off by default and separate from the model request itself.

Hosted Luna

Hosted xgoose offers only Luna. Sign in by email from the extension and make sure your account has available credit. Your extension never receives the service’s upstream provider key.

Trial credit

When the trial campaign is open, verified accounts automatically receive one $5 grant when signing in. It expires 30 days after it is issued. It does not refill daily, and changing devices does not reset eligibility. The platform has a shared daily spending cap of up to $50; service settings can pause claims or inference.

Your account shows granted credit, your available and reserved balances, usage, expiry and platform status. Requests reserve credit before inference; uncertain usage is not treated as zero cost.

Connection endpoints

The protected Hosted API uses POST /ai/v1/chat/completions and GET /ai/v1/models at https://api.xgoose.org. Requests require an active, verified account session. The model list contains Luna only; clients cannot choose an upstream endpoint, key, provider or fallback.

For the normal setup, let the extension manage sign-in, sessions and requests. Keep provider keys in the extension’s connection settings, and never share them in screenshots or public reports.

See the request limits and privacy policy.