Preflight — Deterministic pre-call cost guard for LLM agents. Send a prompt or a message array; get exact o200k_base token counts, per-model input/output cost across 16 frontier models, context-window fit, and the cheapest model that still fits. No LLM call, no upstream API, same answer every time.

POST /v1/preflight   $0.01 USDC (x402)
GET  /v1/price       $0.01 USDC (x402)  ?symbol=BTC
GET  /v1/models      free
GET  /openapi.json   free
GET  /health         free