Getting started

Quickstart

Keep the OpenAI-compatible client you already use, swap the base URL, and verify the economics on one safe slice of traffic first.

5 minFastest production path: create a key, set an env var, swap one base URL.

Base URL swap

Point your existing client at /v1

Workspace key

Create a scoped machine key in the dashboard

Proof headers

Track cache tier, charge, and saved amount

Prerequisites

A Runtime workspace with credits or a connected provider path ready.
A machine key created from the dashboard with the inference scope.
An existing OpenAI-compatible SDK integration you can point at a new base URL.

Integration steps

Create a workspace key

Start in Dashboard / API keys. Runtime keeps human sign-in in the dashboard and expects your app to use a scoped machine key on the live request path.

Set an environment variable

Put the key in GATEWAY_API_KEY so the same secret can be reused across your SDK snippet, your server environment, and quick smoke tests from the terminal.

Swap the base URL

Point the same OpenAI-compatible client at https://api.rntm.sh/v1. That keeps the app contract stable while routing, cache, and billing proof move into Runtime.

1. Create a key

INFO

Recommended first key

Start with an inference-only key for one safe traffic slice. You can add billing or read-only scopes later when a real workflow needs them.

2. Set env var

Authorization header

Authorization: Bearer <GATEWAY_API_KEY>

3. Swap one line

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.GATEWAY_API_KEY,
  baseURL: "https://api.rntm.sh/v1",
});

const response = await client.chat.completions.create({
  model: "gpt-4.1-mini",
  messages: [
    { role: "user", content: "Summarize this support ticket in 3 bullets." },
  ],
});

console.log(response.choices[0]?.message?.content);

Verify it works

The fastest live check is a small request plus an inspection of the response headers. That proves auth, routing, and the savings layer in one shot.

Smoke test

curl -i https://api.rntm.sh/v1/chat/completions \
  -H "Authorization: Bearer $GATEWAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4.1-mini",
    "messages": [{"role":"user","content":"Say hello from Runtime."}]
  }'

Expected response headers

HTTP/2 200
content-type: application/json
x-btl-request-id: req_8f2a1c9d
x-btl-cache-tier: exact_response_cache
x-btl-benchmark-cost: 0.014
x-btl-customer-charge: 0.0056
x-btl-saved: 0.0084
x-btl-net-savings: 0.0084
x-btl-upstream-cost-basis: short_circuited

What's next

Auth

Lock down key scopes

Use scoped machine keys for apps and keep human sign-in inside the dashboard.

Billing proof

Quote before cutover

Run a representative request through the quote route before switching live traffic.

Headers

Read request economics live

Use the response headers and usage APIs to show what Runtime changed on each request.

Was this page helpful?