More

FAQ

Common questions about models, billing and content rules.

Where do I find a model ID?

On the Models page, the small monospace text under each model name is its ID. Hover a row to copy it. GET /models returns the full list.

My client returns 404. What now?

The URL most likely has an extra or missing /v1. Some clients need https://api.geniusrelaygpt.com/v1; others want only the domain. Try the other form.

Why do some models come with and without :thinking?

The :thinking variant reasons before answering. Quality is higher but responses are slower, and reasoning tokens are billed as output. Use the plain variant when you do not need reasoning.

What is a “Private” model?

It runs inside a trusted execution environment, so even the operator cannot see request content. It usually costs a little more than the standard version.

Are there limits on uncensored models?

The models themselves do not refuse, but the Terms of Service still apply: no sexual content involving minors, and no use for fraud, cyberattacks or other illegal activity. Request logs are retained.

Can I get a refund or an invoice?

Redeemed credits cannot be withdrawn. For invoices, contact support.

Is the API reachable from mainland China?

Yes. The endpoint can be reached directly from mainland China without a proxy.