relays · 16 min

AI API relays, explained: how they work and how to audit one

A relay forwards your request to the model provider and returns the answer. That is the whole idea — and every meaningful difference between relays lives in what happens in between.

Supesonic · 30.08.2026 · Developer Library

For live models and prices, use the pricing page and marketplace. Dated examples in the Developer Library are planning aids, not quotes.

An AI API relay is a server that stands between your code and a model provider. You point your SDK at the relay, send the same request you would have sent anyway, and the relay forwards it upstream, gets the answer, and passes it back. Everything that distinguishes one relay from another lives in what happens during that round trip.

The mechanics, in one request

curl https://api.supesonic.com/v1/chat/completions \
  -H "Authorization: Bearer sk-..." \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-4o-mini","messages":[{"role":"user","content":"hi"}]}'
  1. Authentication — the relay checks its own key, not the provider's.
  2. Routing — which upstream credential and endpoint will serve this request.
  3. Forwarding — the request is re-signed with the relay's provider credential.
  4. Accounting — token counts are priced and debit your balance.

How to audit one

  1. GET /v1/models and compare with the marketing catalogue.
  2. Run a canary.
  3. Recompute advertised discounts against provider list prices yourself.
  4. Read the terms and look for a legal entity.
  5. Read the status page sceptically — ask how many attempts and samples sit behind each figure.

Supesonic is an OpenAI-compatible relay. Claims are not evidence — run the checks yourself, including against us.

← Developer Library