> ## Documentation Index
> Fetch the complete documentation index at: https://docs.brightnode.cloud/llms.txt
> Use this file to discover all available pages before exploring further.

# Router

> Use Brightnode's OpenAI-compatible inference endpoint with one API key and one base URL.

Router is Brightnode's hosted inference entrypoint. It gives you a single OpenAI-compatible API for the models Brightnode exposes through the shared catalog, so you can swap models without changing SDKs or reworking your application.

<Note>
  Use `https://api.brightnode.cloud/v1` as the base URL for Router requests.
</Note>

## Quickstart

Before you send traffic, create an [API key](/get-started/api-keys) with the `Inference` scope.

### List available models

```bash theme={"system"}
curl https://api.brightnode.cloud/v1/models \
  -H "Authorization: Bearer $BRIGHTNODE_API_KEY"
```

### Send a chat completion

```bash theme={"system"}
curl https://api.brightnode.cloud/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BRIGHTNODE_API_KEY" \
  -d '{
    "model": "meta-llama/Llama-3.3-70B-Instruct",
    "messages": [
      { "role": "user", "content": "Give me three ideas for a launch email." }
    ],
    "max_tokens": 256
  }'
```

### Generate embeddings

```bash theme={"system"}
curl https://api.brightnode.cloud/v1/embeddings \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $BRIGHTNODE_API_KEY" \
  -d '{
    "model": "Qwen/Qwen3-Embedding-8B",
    "input": "Brightnode hosted inference"
  }'
```

## Use Router with OpenAI SDKs

Router is designed to be drop-in compatible with OpenAI SDKs when you set the base URL to `https://api.brightnode.cloud/v1`.

<CodeGroup>
  ```python Python theme={"system"}
  from openai import OpenAI
  import os

  client = OpenAI(
      base_url="https://api.brightnode.cloud/v1",
      api_key=os.environ["BRIGHTNODE_API_KEY"],
  )

  response = client.chat.completions.create(
      model="meta-llama/Llama-3.3-70B-Instruct",
      messages=[{"role": "user", "content": "Hello from Brightnode"}],
      max_tokens=128,
  )

  print(response.choices[0].message.content)
  ```

  ```typescript TypeScript theme={"system"}
  import OpenAI from "openai";

  const client = new OpenAI({
    baseURL: "https://api.brightnode.cloud/v1",
    apiKey: process.env.BRIGHTNODE_API_KEY,
  });

  const response = await client.chat.completions.create({
    model: "meta-llama/Llama-3.3-70B-Instruct",
    messages: [{ role: "user", content: "Hello from Brightnode" }],
    max_tokens: 128,
  });

  console.log(response.choices[0].message.content);
  ```
</CodeGroup>

## Request behavior

Router supports the standard hosted inference paths:

* `GET /v1/models` to list available models.
* `GET /v1/models/{model_id}` to inspect a single model.
* `POST /v1/chat/completions` for chat-style generation.
* `POST /v1/completions` for legacy text completion clients.
* `POST /v1/embeddings` for embedding workloads.

If a requested model is still waking up, Router may respond with `503` and a `Retry-After` header. In that case, retry after the suggested delay.

## When to use Router

Use Router when you want:

* One endpoint across multiple hosted models.
* Standard OpenAI SDK compatibility.
* Centralized API key management.
* Inference analytics in the Brightnode console.

If you need deployment-level control instead of the shared hosted endpoint, see [Beams](/beams/overview).