Frequently Asked Questions
Answers to common questions regarding billing, latency, BYOK, and enterprise compliance.
Frequently Asked Questions
Does Zorveus add latency to AI inference?
No. The Zorveus Gateway is built on high-performance async streaming infrastructure that adds less than 5ms overhead. Token streams pass directly through Server-Sent Events (SSE) with zero response buffering.
Can I mix BYOK keys and wallet-funded routing?
Yes. If you add an OpenAI BYOK credential with model policy openai/*, all OpenAI requests route through your direct key. If you then call anthropic/claude-3-7-sonnet without an Anthropic key, Zorveus automatically falls back to your funded organization wallet.
Are prompt inputs and completions stored?
No. Zorveus operates a zero-data-retention policy for model inputs and completions by default. Only metadata (token counts, timestamps, user ID, cost) is stored in the ledger.
Was this page helpful?
Edit this page on GitHub