Connect NordAI to your app

The most common failure: API Type = “Responses API”. NordAI implements the Chat Completions protocol only. A panel set to Responses API probes POST /v1/responses, an endpoint NordAI does not have, and reports the false error “API key unauthorized” even though the key is fine. Always pick Chat Completions (some panels label it “OpenAI Chat” or just “OpenAI”).

Name:     NordAI
Prefix:   nordai            # model ids become nordai/nordai-claude-opus-5 in your panel
API Type: Chat Completions  ← NOT "Responses API"
Base URL: https://api.nordai.id/v1     (must end in /v1)
API Key:  nord_sk_…                    (a key from dashboard/api-keys, not a device token)
Model ID: nordai-claude-opus-5         (optional; or leave empty and import from /v1/models)

Checklist when “Check” fails

• “API key unauthorized” — 90% of the time this is API Type = Responses API. Switch to Chat Completions and re-check. If it still fails, verify the key starts with nord_sk_, was copied whole (no spaces), and is still active on the API keys page.
• 404 / “model not found” during check — the Model ID field must be a NordAI id such as nordai-claude-opus-5, not gpt-4. Get valid ids from GET https://api.nordai.id/v1/models.
• Connection timeout — some panels reject self-signed or unusual TLS; NordAI uses a standard Cloudflare certificate, so a timeout usually means the Base URL is missing /v1 or the panel cannot reach the internet.
• Model list is empty — import models manually: ids follow nordai-<name> (e.g. nordai-claude-opus-5, nordai-deepsek-4-1-flash). The full catalogue is at GET /v1/models.

After the channel is added

Run one test chat from the panel on a NordAI model. Expect HTTP 200 and content in choices[0].message.content. Then the request shows up within seconds in Dashboard Usage — that is the confirmation that routing, key and billing are all live end to end.

Provider:   OpenAI-compatible / Custom
Base URL:   https://api.nordai.id/v1
API Key:    nord_sk_…
Model:      nordai-claude-opus-5

import os
from litellm import completion

os.environ["OPENAI_API_KEY"] = "nord_sk_…"

resp = completion(
    model="openai/nordai-claude-opus-5",
    api_base="https://api.nordai.id/v1",
    messages=[{"role": "user", "content": "hi"}],
)

The openai/ prefix tells LiteLLM to speak the OpenAI protocol against your api_base.

# Python
from openai import OpenAI
client = OpenAI(base_url="https://api.nordai.id/v1", api_key="nord_sk_…")

# Node
import OpenAI from "openai";
const client = new OpenAI({ baseURL: "https://api.nordai.id/v1", apiKey: "nord_sk_…" });

const r = await client.chat.completions.create({
  model: "nordai-claude-opus-5",
  messages: [{ role: "user", content: "hi" }],
});

import { createOpenAI } from "@ai-sdk/openai";

const nordai = createOpenAI({
  baseURL: "https://api.nordai.id/v1",
  apiKey: process.env.NORDAI_API_KEY,
});

await streamText({ model: nordai("nordai-claude-opus-5"), prompt: "…" });

Use the built-in OpenAI node, set "Base URL" (or "Override URL") to https://api.nordai.id/v1, paste your nord_sk_… key as the credential, and type the model id manually. No-code tools that hide the base URL need an HTTP Request node hitting https://api.nordai.id/v1/chat/completions directly with the bearer header.