---
title: "OpenAI Gateway"
description: "Connect directly to the OpenAI API"
---

The OpenAI gateway connects ChatJS directly to the [OpenAI API](https://platform.openai.com/docs/api-reference) using the `@ai-sdk/openai` provider SDK. Use this when you only need OpenAI models and want the simplest setup with type-safe model IDs.

## Setup

1. Follow the [OpenAI quickstart](https://developers.openai.com/api/docs/quickstart) to create an API key
2. Add to `.env.local`:

```bash
OPENAI_API_KEY=sk-...
```

3. Set the gateway in your config:

```typescript title="chat.config.ts"
const config: ConfigInput = {
  ai: {
    gateway: "openai",
    workflows: {
      chat: "gpt-5-mini",
      title: "gpt-5-nano",
      // ...
    },
    // ...
  },
};
```

## Authentication

| Variable         | Description                   |
| ---------------- | ----------------------------- |
| `OPENAI_API_KEY` | Required. Your OpenAI API key |

## Available Models

All models available to your API key (GPT, DALL-E, and embedding models). The full list is fetched at runtime and cached for 1 hour.

## Model IDs

Unlike the Vercel and OpenRouter gateways which use `provider/model` format, the OpenAI gateway uses bare model IDs since all models come from OpenAI:

- `gpt-5-mini`
- `gpt-5-nano`
- `gpt-4o`
- `o3-mini`

The gateway provides **type-safe model IDs** extracted from the `@ai-sdk/openai` SDK, giving you autocomplete in `chat.config.ts`.

## Image Generation

The OpenAI gateway supports dedicated image models through `createImageModel()`:

- `gpt-image-1`
- `gpt-image-1.5`
- `dall-e-3`

This enables the built-in [Image Generation](/features/image-generation) feature.

## Related

- [Gateways Overview](/gateways/overview) for the full gateway comparison
- [Vercel Gateway](/gateways/vercel) for multi-provider access via a single key