Billing & credits
Hormouz bills with a hold-and-settle model: before a request reaches the model, its estimated cost is held from your credit; after it finishes, the actual cost is settled and the difference released. If your balance can't cover the hold, the request is rejected upfront with 402 INSUFFICIENT_CREDITS and nothing is charged.
Safe retries with Idempotency-Key
Send an Idempotency-Key header on POST requests (any string up to 200 chars). The same key with the same body replays the cached response. No re-run, no double charge. The same key with a different body returns 409 CONFLICT. Keys expire after 24 hours.
Where to see each request's cost
Embeddings return the cost inline in usage.cost; in streamed chat, the final chunk before [DONE] carries the settled cost. Your full usage history and invoices live in the Hormouz dashboard.