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

# BYOK Setup

> Create and manage custom provider keys (BYOK)

Create and manage custom provider keys (BYOK - Bring Your Own Key) for AWS Bedrock, Google Vertex AI, Azure, and other LLM providers. This allows you to use your own credentials instead of Edgee's shared keys.

## Request Body

<ParamField body="provider" type="string" required>
  The provider type. Supported values: `anthropic`, `openai`, `google_vertex_ai`, `mistral`, `deepseek`, `xai`, `zai`, `bedrock`, `azure`
</ParamField>

<ParamField body="provider_key" type="string" required>
  The credentials for the provider. Format depends on the provider type:

  * **Standard providers** (anthropic, openai, mistral, deepseek, xai, zai): API key as string
  * **Bedrock**: JSON string with region-keyed credentials
  * **Azure**: JSON string with endpoint and api\_key
  * **Google Vertex AI**: Service account JSON string
</ParamField>

<ParamField body="name" type="string" required>
  A descriptive name for this provider key (e.g., "Production Vertex AI Key"). Max 100 characters.
</ParamField>

<ParamField body="api_key_ids" type="array" required>
  Array of Edgee API key IDs to assign this provider key to. Leave empty to create an organization-level provider key that applies to all API keys.
</ParamField>

## Response

<ResponseField name="id" type="string">
  Unique identifier for the provider key
</ResponseField>

<ResponseField name="provider" type="string">
  The provider type (e.g., "google\_vertex\_ai")
</ResponseField>

<ResponseField name="name" type="string">
  The name of the provider key
</ResponseField>

<ResponseField name="created_at" type="string">
  ISO 8601 timestamp of when the provider key was created
</ResponseField>

<ResponseField name="created_by" type="string">
  User ID of who created the provider key
</ResponseField>

## Examples

<RequestExample>
  ```bash Google Vertex AI theme={"dark"}
  curl -X POST https://api.edgee.app/v1/organizations/{orgId}/provider-keys \
    -H "Authorization: Bearer <your_api_token>" \
    -H "Content-Type: application/json" \
    -d '{
      "provider": "google_vertex_ai",
      "provider_key": "{\"type\":\"service_account\",\"project_id\":\"my-project\",\"private_key\":\"...\"}",
      "name": "Production Vertex AI",
      "api_key_ids": []
    }'
  ```

  ```bash AWS Bedrock (Multi-region) theme={"dark"}
  curl -X POST https://api.edgee.app/v1/organizations/{orgId}/provider-keys \
    -H "Authorization: Bearer <your_api_token>" \
    -H "Content-Type: application/json" \
    -d '{
      "provider": "bedrock",
      "provider_key": "{\"global\":{\"access_key_id\":\"AKIA...\",\"secret_access_key\":\"...\"},\"us-east-1\":{\"access_key_id\":\"AKIA...\",\"secret_access_key\":\"...\"}}",
      "name": "AWS Bedrock Production",
      "api_key_ids": []
    }'
  ```

  ```bash Azure theme={"dark"}
  curl -X POST https://api.edgee.app/v1/organizations/{orgId}/provider-keys \
    -H "Authorization: Bearer <your_api_token>" \
    -H "Content-Type: application/json" \
    -d '{
      "provider": "azure",
      "provider_key": "{\"endpoint\":\"https://your-resource.openai.azure.com/\",\"api_key\":\"your-api-key\"}",
      "name": "Azure OpenAI",
      "api_key_ids": []
    }'
  ```

  ```bash With API Key Assignment theme={"dark"}
  curl -X POST https://api.edgee.app/v1/organizations/{orgId}/provider-keys \
    -H "Authorization: Bearer <your_api_token>" \
    -H "Content-Type: application/json" \
    -d '{
      "provider": "openai",
      "provider_key": "sk-proj-...",
      "name": "OpenAI - Development",
      "api_key_ids": ["key-id-1", "key-id-2"]
    }'
  ```
</RequestExample>

<ResponseExample>
  ```json theme={"dark"}
  {
    "id": "pk_1a2b3c4d5e6f7g8h",
    "provider": "google_vertex_ai",
    "name": "Production Vertex AI",
    "created_at": "2026-04-08T10:00:00Z",
    "created_by": "user_abc123"
  }
  ```
</ResponseExample>

## Tips

* **Organization-level keys**: Leave `api_key_ids` empty to apply the provider key to your entire organization
* **API-key-specific assignments**: Specify API key IDs to restrict the provider key to specific API keys only
* **Security**: Never commit provider credentials to version control. Use environment variables or secure credential management systems
* **Multi-region Bedrock**: For AWS Bedrock, you can specify different credentials for different regions. Use `"global"` as the key for a default region.

## Credential Formats

### Google Vertex AI

Provide the entire service account JSON as a string:

```json theme={"dark"}
{
  "type": "service_account",
  "project_id": "your-project-id",
  "private_key_id": "key-id",
  "private_key": "-----BEGIN PRIVATE KEY-----\n...\n-----END PRIVATE KEY-----\n",
  "client_email": "service-account@your-project.iam.gserviceaccount.com",
  "client_id": "123456789",
  "auth_uri": "https://accounts.google.com/o/oauth2/auth",
  "token_uri": "https://oauth2.googleapis.com/token"
}
```

### AWS Bedrock

Provide region-keyed credentials:

```json theme={"dark"}
{
  "global": {
    "access_key_id": "AKIA...",
    "secret_access_key": "..."
  },
  "us-east-1": {
    "access_key_id": "AKIA...",
    "secret_access_key": "..."
  }
}
```

### Azure

Provide endpoint and API key:

```json theme={"dark"}
{
  "endpoint": "https://your-resource.openai.azure.com/",
  "api_key": "your-api-key"
}
```

<EdgeeSdk />