Frequently asked questions

An OpenAI- and Anthropic-compatible gateway in front of a pool of upstream AI providers. One key reaches every model on the pricing page.

27 model IDs are live right now. The exact, case-sensitive list comes from GET /api/v1/models; see the pricing page.

Usage is metered per token and charged against your balance. Input and output tokens are charged at the same per-model rate. Deposits are USDC on Solana, credited 1 USDC = 1,000,000 metering units.

Dashboard → Deposit. Each account gets a unique Solana address. Send USDC to it; the balance is credited after the transfer confirms on-chain. Double-check the network — funds sent elsewhere cannot be recovered.

See the refund policy. In short: unused balance is not automatically refundable, and mistakes on your side (wrong network, wrong address) are out of our control.

There is no per-key request-per-minute cap. Capacity is bounded by your token balance and by a concurrency ceiling on the gateway, so a burst can be queued or rejected when the ceiling is reached.

Account passwords are stored as bcrypt hashes. API keys are stored as SHA-256 hashes with only a short prefix kept for display, which is why a key is shown exactly once. Upstream credentials you attach as a publisher are stored encrypted at rest.

No. Prompts and completions are passed through to the selected upstream and recorded for usage metering only.

Yes — open a request on Discord and the account and its keys are removed.