Quickstart

Relay speaks the OpenAI wire format. Change the base URL and the key; leave your request bodies alone. The Anthropic-compatible endpoint is built but not live yet — see Not live yet below.

Create a key under API keys. The secret is shown exactly once. Every key on your account draws down the same prepaid balance.

The OpenAI SDK below sends Authorization: Bearer <key> for you — just set api_key. Model ids are the short ids listed on Models (glm-5.3, not z-ai/glm-5.3); anything not on that page is not callable yet.

curl

curl https://ai-gateway.getborg.com/api/v1/proxy/v1/chat/completions \
  -H "Authorization: Bearer $RELAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm-5.3",
    "messages": [{"role": "user", "content": "In one line: what is prepaid inference?"}]
  }'

OpenAI SDK — Python

from openai import OpenAI

client = OpenAI(
    base_url="https://ai-gateway.getborg.com/api/v1/proxy/v1",
    api_key=os.environ["RELAY_API_KEY"],
)

resp = client.chat.completions.create(
    model="glm-5.3",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

OpenAI SDK — JavaScript

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://ai-gateway.getborg.com/api/v1/proxy/v1",
  apiKey: process.env.RELAY_API_KEY,
});

const resp = await client.chat.completions.create({
  model: "gemini-3-flash",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(resp.choices[0].message.content);

Cursor

Settings → Models → OpenAI API Key: tick Override OpenAI Base URL, set it to https://ai-gateway.getborg.com/api/v1/proxy/v1, and paste your Relay key as the API key. Add glm-5.3 as a custom model — Cursor's built-in model names are GPT and Claude ids, which are not on sale yet.

Not live yet

We would rather tell you than let you debug it. These are built but not callable on a Relay key today, so they are not priced on Models and there is no example for them here:

• GPT, Claude, Grok, Muse and Kimi models. Capacity and entitlement work in progress. Use glm-5.3 or gemini-3-flash meanwhile — same OpenAI-compatible request body, just a different "model".
• The Anthropic-compatible endpoint (and so Claude Code via ANTHROPIC_BASE_URL), because it needs a Claude-family model.
• Codex CLI, which pins a GPT model id.
• Image generation, not a live route yet.

Everything above is on the way; nothing here costs you credit in the meantime. Ask us if a specific model is blocking you and we will tell you where it stands.

Errors worth knowing