openai-compatible · one endpoint · x402 settlement

The inference layer that pays for itself.

Every prompt routed to the cheapest capable model, failed over when a provider blinks, billed per call in USDC from your wallet — the same answer for a smaller bill. Each reply shows exactly what it cost and what you saved.

Wallet-funded · routed to the cheapest capable model · metered to the token
decisions System One · state in → typed answers out · metered to the token

Decisions, not strings

Describe any state — a ticket, a record, a paragraph — and ask typed questions. Every answer comes back as a typed value with calibrated probabilities and confidence your code can branch on. Billed per call from your wallet.

history Your chat threads, newest first
activity Every metered event — replies, decisions, settlements
guardrails Safety + quality judgments over every reply
settings This deployment, at a glance