# OpenAI SDK

The
[OpenAI Node.js SDK](https://platform.openai.com/docs/libraries/node-js-library)
is the official SDK for working with LLMs provided by OpenAI as well as other
OpenAI compatible models from other providers in Node.js.

## Prerequisites

In order to use the AI Gateway with any OpenAI SDK powered application you will
need to complete these steps first:

<Stepper>

1. Create a [new provider](/ai-gateway/managing-providers) in the AI Gateway for
   OpenAI

2. [Set up a new team](/ai-gateway/managing-teams)

3. Create a [new app](/ai-gateway/managing-apps) to use specifically with the
   OpenAI SDK and assign it to the team you created

4. Copy the API Key for the app you created, as well as the Gateway URL

</Stepper>

## Configure the OpenAI SDK

To route all OpenAI SDK requests through Zuplo instead of directly to the OpenAI
API, you must set the API `baseUrl` in the SDK configuration.

For example, if your Zuplo application is hosted at
https://my-ai-gateway.zuplo.app, you can change the base URL in your API client
to https://my-ai-gateway.zuplo.app/v1.

Additionally, change the value of `apiKey` to the API key of the app you have
configured in Zuplo

```typescript
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.ZUPLO_AI_GATEWAY_API_KEY,
  baseURL: "https://my-ai-gateway.zuplo.app/v1",
});

const response = await client.chat.completions.create({
  model: "gpt-4",
  messages: [
    {
      role: "user",
      content: "Write a one-sentence bedtime story about a unicorn.",
    },
  ],
});

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

When configured in this way, the SDK will switch from using the default OpenAI
APIs to using Zuplo's AI Gateway for all requests.

## Supported Endpoints

The AI Gateway supports all major OpenAI API endpoints through the universal
API:

- **Chat Completions** (`/v1/chat/completions`) - For conversational AI
  interactions
- **Embeddings** (`/v1/embeddings`) - For generating vector embeddings
- **Responses** (`/v1/responses`) - For OpenAI models that support the responses
  endpoint (such as GPT-5 and other compatible models)

## Using the Responses Endpoint

For models that support the `/v1/responses` endpoint, you can use the OpenAI
SDK's `responses` API:

```typescript
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.ZUPLO_AI_GATEWAY_API_KEY,
  baseURL: "https://my-ai-gateway.zuplo.app/v1",
});

const response = await client.responses.create({
  model: "gpt-5",
  input: "Write a one-sentence bedtime story about a unicorn.",
});

console.log(response.output_text);
```

The responses endpoint is automatically routed through the AI Gateway, providing
the same monitoring, cost controls, and security features as other endpoints.