> ## Documentation Index
> Fetch the complete documentation index at: https://developer.watson-orchestrate.ibm.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Enabling watsonx/* LLMs

The **watsonx Orchestrate Developer Edition** now supports integration with the **wxO SaaS tenant** for LLM inferencing. Previously, developers were required to configure the Developer Edition using their own `watsonx.ai` API key and Space ID. This update simplifies the setup process by enabling proxy usage through Watsonx Orchestrate.

This configuration approach also applies to other supported LLM providers, including **Groq** and **AWS Bedrock**.

## Prerequisites

To retrieve the necessary configuration details, you must have access to an instance of **watsonx Orchestrate**.

## Environment Configuration

Additional environment variables are required to enable proxy usage. These variables differ based on the deployment type:

* [**SaaS Deployment**](#saas-deployment-configuration)
* [**On-premises Deployment**](#on-premises-deployment-configuration)

### Common Parameters

| Variable                    | Description                                                                |
| --------------------------- | -------------------------------------------------------------------------- |
| `AUTHORIZATION_URL`         | IAM URL used to authorize watsonx Orchestrate to access the proxy service. |
| `USE_SAAS_ML_TOOLS_RUNTIME` | Enables the ADK to use the proxy.                                          |

### SaaS Deployment Configuration

For SaaS deployments, use the following environment variables:

```env theme={null}
WO_API_KEY=<your API key for Watson Orchestrate on SaaS>
WO_INSTANCE=<your service instance URL>
# The below two fields are optional and automatically inferred by the Developer Edition
AUTHORIZATION_URL=<your IAM URL>
USE_SAAS_ML_TOOLS_RUNTIME=True
```

<Note>
  **Note:**\
  `watsonx.ai` API Key and Space ID are no longer required.
</Note>

### On-premises Deployment Configuration

For on-premises deployments, use the following environment variables:

<Tabs>
  <Tab title="authenticating with password">
    ```env theme={null}
    WO_USERNAME=<your username for onprem>
    WO_PASSWORD=<your password for onprem>
    WO_INSTANCE=<your service instance URL>
    AUTHORIZATION_URL=<your IAM URL>
    ```

    | Variable            | Description                                                                             |
    | ------------------- | --------------------------------------------------------------------------------------- |
    | `WO_USERNAME`       | Username used to log in to IBM SoftwareHub.                                             |
    | `WO_PASSWORD`       | Password associated with `WO_USERNAME`.                                                 |
    | `WO_INSTANCE`       | watsonx Orchestrate server instance URL                                                 |
    | `AUTHORIZATION_URL` | Optional, an IAM URL used to authorize watsonx Orchestrate to access the proxy service. |
  </Tab>

  <Tab title="authenticating with API key">
    ```env theme={null}
    WO_USERNAME=<your username for onprem>
    WO_API_KEY=<your API key for onprem>
    WO_INSTANCE=<your service instance URL>
    AUTHORIZATION_URL=<your IAM URL>
    ```

    | Variable            | Description                                                                             |
    | ------------------- | --------------------------------------------------------------------------------------- |
    | `WO_USERNAME`       | Username used to log in to IBM SoftwareHub.                                             |
    | `WO_API_KEY`        | API key associated with `WO_USERNAME`.                                                  |
    | `WO_INSTANCE`       | watsonx Orchestrate server instance URL                                                 |
    | `AUTHORIZATION_URL` | Optional, an IAM URL used to authorize watsonx Orchestrate to access the proxy service. |
  </Tab>
</Tabs>

## Alternative LLM Providers

In addition to [`watsonx.ai`](https://www.ibm.com/watsonx) models, the Developer Edition supports other LLM providers. Similar to watsonx.ai configuration, you can either provide your own credentials or use the watsonx Orchestrate proxy service.

### Groq Configuration

To use the [`groq/openai/gpt-oss-120b`](https://groq.com/) model, you have two options:

**Option 1: Using Your Own Groq Credentials**

Add the following to your `.env` file:

```env theme={null}
GROQ_API_KEY=<your-groq-key>
```

**Option 2: Using watsonx Orchestrate Proxy**

Configure the same `WO_API_KEY` and `WO_INSTANCE` variables as described in the [SaaS Deployment Configuration](#saas-deployment-configuration) or [On-premises Deployment Configuration](#on-premises-deployment-configuration) sections above.

### AWS Bedrock Configuration

To use the [`bedrock/openai.gpt-oss-120b-1:0`](https://aws.amazon.com/bedrock/) model, you have two options:

**Option 1: Using Your Own AWS Bedrock Credentials**

Add the following to your `.env` file:

```env theme={null}
BEDROCK_AWS_ACCESS_KEY_ID=<your-bedrock-access-key-id>
BEDROCK_AWS_SECRET_ACCESS_KEY=<your-bedrock-secret-access-key>
```

**Option 2: Using watsonx Orchestrate Proxy**

Configure the same `WO_API_KEY` and `WO_INSTANCE` variables as described in the [SaaS Deployment Configuration](#saas-deployment-configuration) or [On-premises Deployment Configuration](#on-premises-deployment-configuration) sections above.

<Note>
  **Note:**
  When using the watsonx Orchestrate proxy (Option 2), you don't need to provide provider-specific credentials. The proxy handles authentication with the LLM providers on your behalf.
</Note>
