To ensure that only authenticated clients can access your AI models, you must deploy an inference instance with authorization enabled.

​
Step 1. Enable authorization

When deploying an AI model, set the auth_enabled option to true. This means an API Key will be automatically generated and linked to the deployment.

​
Step 2. Retrieve the API key

Once the deployment is created with authentication enabled, you can retrieve the API Key via the designated API endpoint.

​
API request

The API key can be retrieved via this endpoint.
InfoThe API Key is only available through this endpoint. Store it securely.

​
Step 3. Use the API key for authorization

Once you have retrieved the API Key, include it in your API requests using the X-API-Key header.

​
Example using OpenAI python client library

Here’s an example demonstrating how to use the API key for authorization: 
from openai import OpenAI

def get_llm_response(message: str) -> str:
    client = OpenAI(api_key=LLM_KEY, base_url=f"{LLM_API}/v1")

    response = client.chat.completions.create(
        model="meta-llama/Llama-3.3-70B-Instruct",
        messages=[
            {"role": "user", "content": message},
        ],
        extra_headers={"X-API-Key": LLM_KEY},
    )
    return response.choices[0].message.content

if __name__ == "__main__":
    print(get_llm_response("Why is the sky blue?"))
InfoFor Gcore deployments with authorization enabled, the X-API-Key header is mandatory in all API requests.
To learn more about deploying AI models, refer to our dedicated guide.