Direct model provider

Use OpenAI through Relays.

Connect OpenAI models to Relays while keeping Chat Completions and native Responses requests at the edge.

Talk to us Browse every provider

One alias in front of OpenAI.

Your application calls a Relays alias. Relays holds the OpenAI credential, applies the route policy, and hands back the provider response in the shape your SDK expects.

  1. CallerYour application

    Keeps the OpenAI request it already sends.

  2. EdgeRelays

    Holds the key, picks the candidate, writes the receipt.

  3. UpstreamOpenAI

    Receives the call on its own base URL.

Wire protocol
OpenAI
Relays endpoints
  • Chat Completions
  • Responses
Credential boundary
Bring your own OpenAI API keyOPENAI_API_KEY
Upstream base URL
https://api.openai.com/v1
Connection
Registry entry
Source
OpenAI API documentation

Relays keeps the OpenAI request family intact and forces Responses store:false for its stateless gateway boundary.

What is measured, and what is not.

Availability comes from OpenAI. Latency, error rate, and cost per successful request come from your own traffic, so Relays leaves them empty until requests run.

Upstream status

OpenAI

Operational

Chat Completions
Operational
Responses
Operational

Relays measurements

Your traffic
No sample yet
Time to first token, error rate, and cost per successful request are read from requests that ran through Relays. This route has none, so there is nothing to plot.
Ask for an API key

Models at the edge.

Relays keeps a copy of the OpenAI catalogue and the endpoint each model is routed to. Pick one to read what it is for.

16 in catalogue · 4 routed

Selected model
gpt-5.6-sol
Built for
Complex reasoning and coding
Provider family
Frontier
Relays edge
Chat Completions · Responses

Catalogue read from OpenAI model catalog · Aug 7, 2026

Every route carries a policy.

A route is a list of candidates, a budget, and a record of what happened. You set all three before any traffic runs.

Fallback

Candidates stay inside the contract.

Chat Completions and Responses routes can use ordered fallback candidates that declare the same Relays edge endpoint. Relays does not translate Responses into another API family to make a fallback fit.

See how a route is chosen
Budget

A ceiling on this route.

Set an alert threshold and a stop threshold for the route rather than for the whole application. Relays checks both before the call goes upstream.

Receipt

Each attempt is written down.

Every call records the alias, the candidates tried, the status each returned, tokens in and out, and the estimated cost of the attempt that served it.

Read a route receipt

One endpoint is enough to start.

Keep the client you already wrote. Bring us the workload and we will map the first OpenAI route with you.

Start a conversation