Drop one base_url into your OpenAI SDK. Send claude-opus-4-8, gpt-5.5, gemini-3.1-pro, or anything else worth calling. Same key, same shape, every model.
A single OpenAI-compatible router. You send one shape of request, with a model id, and it lands on a model that can serve it — no matter whose backend that is.
Built for builders who change models the way they change libraries. The same key works for claude-opus-4-8, gpt-5.5, gemini-3.1-pro, deepseek-r1, grok-4.3, and the image families.
OpenAI shape on the wire. Drop into the openai SDK by changing base_url.
Bearer hr-… you mint yourself. No middlemen, no third-party login.
Pay-as-you-go from a balance you top up. No subscriptions, no card to start.
Replace api.openai.com with api.holyrouter.xyz. Drop in your key. Send the same request shape. That is the whole migration.
curl https://api.holyrouter.xyz/v1/chat/completions \
-H "Authorization: Bearer hr-…" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-opus-4-8",
"messages": [{"role":"user","content":"hi"}]
}' from openai import OpenAI
client = OpenAI(
base_url="https://api.holyrouter.xyz/v1",
api_key="hr-…",
)
resp = client.chat.completions.create(
model="gpt-5.5",
messages=[{"role": "user", "content": "hi"}],
)
print(resp.choices[0].message.content) import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.holyrouter.xyz/v1",
apiKey: "hr-…",
});
const r = await client.chat.completions.create({
model: "gemini-3.1-pro",
messages: [{ role: "user", content: "hi" }],
});
console.log(r.choices[0].message.content); Yes. Sign in, get a key, $10 lands in your balance — no card, no trial expiry, no auto-charge. When the balance runs out, requests stop. Top up if you want more; walk away if you don't.
One OpenAI-compatible endpoint that resolves your model id to a working upstream, handles authentication, billing and retries, and gives you back the same chat / images shape you already know. Change one URL and one key — your code stays the same.
Claude opus-4-8 and sonnet-4-5, GPT 5.5 (and 5.5-mini), Gemini 3.1-pro and 3.5-flash, Grok 4.3, DeepSeek r1 and v3, Qwen3, plus image families like flux-pro, seedream, ideogram. One key, one bill.
If you only ever call OpenAI, you can. The moment you want a second model — claude, gemini, grok — holyrouter pays for itself: same SDK, one balance, one failover path.
By the model id you send. We never silently swap models. If the requested model can serve, your request lands on it. If not, you get an honest error — not a different model dressed up.