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

# Omnifact Provider

> Create and configure the Omnifact provider for chat completions via the API Gateway

## Overview

The Omnifact provider connects to the [Omnifact API Gateway](https://connect.omnifact.ai) — an OpenAI-compatible endpoint for chat completions. Use your Omnifact organization API key and any model enabled for your organization.

## Installation

```bash theme={null}
npm install @core-ai/omnifact
```

## createOmnifact()

Create an Omnifact provider instance.

```typescript theme={null}
import { createOmnifact } from '@core-ai/omnifact';

const omnifact = createOmnifact({
  apiKey: process.env.OMNIFACT_API_KEY,
});
```

### Options

<ParamField path="apiKey" type="string" optional>
  Your Omnifact organization API key.
</ParamField>

<ParamField path="baseURL" type="string" optional>
  Custom gateway base URL. Defaults to `https://connect.omnifact.ai/v1/gateway`.
</ParamField>

### Returns

`OmnifactProvider` with method `chatModel()`.

## Model IDs

Use the exact `id` values returned by `GET /v1/gateway/models`. The SDK passes the model id through unchanged.

EU-hosted backends (Azure OpenAI, Vertex Google, Vertex Anthropic, Mistral, etc.) always use the `eu/` prefix:

* `eu/gpt-5-mini` — Azure OpenAI
* `eu/claude-4-6-sonnet` — Vertex Anthropic
* `eu/mistral-large` — Mistral

Direct non-EU providers use the plain type id:

* `gpt-5-mini` — OpenAI
* `claude-4-6-sonnet` — Anthropic

When both a direct and an EU backend exist for the same model type, both ids may appear in the model list. Use the one that matches where you want the request routed.

## Capabilities

| Feature          | Support          |
| ---------------- | ---------------- |
| Chat Completion  | Yes              |
| Streaming        | Yes              |
| Function Calling | No               |
| Vision           | No               |
| Reasoning        | Depends on model |
| Embeddings       | No               |
| Image Generation | No               |

## Examples

### Basic chat

```typescript theme={null}
import { createOmnifact } from '@core-ai/omnifact';
import { generate } from '@core-ai/core-ai';

const omnifact = createOmnifact({
  apiKey: process.env.OMNIFACT_API_KEY,
});

// Use an id from GET /v1/gateway/models (eu/ prefix for EU-hosted models).
const model = omnifact.chatModel('eu/gpt-5-mini');

const result = await generate({
  model,
  messages: [{ role: 'user', content: 'Hello!' }],
});

console.log(result.content);
```

### Streaming

```typescript theme={null}
import { createOmnifact } from '@core-ai/omnifact';
import { stream } from '@core-ai/core-ai';

const omnifact = createOmnifact({
  apiKey: process.env.OMNIFACT_API_KEY,
});

// Use an id from GET /v1/gateway/models (eu/ prefix for EU-hosted models).
const model = omnifact.chatModel('eu/gpt-5-mini');

const result = await stream({
  model,
  messages: [{ role: 'user', content: 'Count to five.' }],
});

for await (const chunk of result.textStream) {
  process.stdout.write(chunk);
}
```

## Provider-specific options

The gateway follows the OpenAI Chat Completions API. Pass options via `providerOptions.openai`:

```typescript theme={null}
await generate({
  model: omnifact.chatModel('eu/gpt-5-mini'),
  messages: [{ role: 'user', content: 'Hello!' }],
  providerOptions: {
    openai: {
      temperature: 0.7,
    },
  },
});
```

See the [OpenAI compat provider](/api/providers/openai#createopenaicompat) docs for available Chat Completions options.