flatfee.one Docs

Quickstart

Use the same chat completions shape most OpenAI-compatible tools already support.

curl https://api.flatfee.one/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "auto",
    "messages": [{
      "role": "user",
      "content": "Run this coding task through flatfee.one."
    }]
  }'

Authentication

Send the API key as a bearer token. Do not expose keys in browser clients.

HeaderAuthorization: Bearer YOUR_API_KEY

Content typeapplication/json

Key scopeProject keys during private beta. Team keys are planned.

Model selection

auto lets the gateway choose the lane. Explicit model aliases will be documented before public launch.

autoUse routing policy for premium, spend-control, and continuation lanes.

premiumPrefer stronger reasoning capacity when available.

economyPrefer lower-cost continuation behavior.

Exact model/provider availability can change during beta. See Models for the public policy.

Headers

Response headers are planned as the visible receipt for routing and limits.

x-flatfee-lanepremium, budget, continuation, or fallback.

x-flatfee-fair-use-remainingPremium fair-use estimate for the current billing window.

x-flatfee-request-idSupport and trace correlation.

Errors

Named errors make agent behavior easier to debug than a generic provider failure.

401Missing or invalid API key.

402Plan or billing state prevents premium routing.

429Concurrent workflow or abuse-control limit reached.

503Provider connectivity or gateway capacity issue.

fallback_exhaustedAll eligible fallback routes failed or were unavailable.

Billing behavior

Plans include premium fair-use and documented continuation. The point is predictable spend, not pretending compute has no limits.