How to access the OpenAI 2024-11-20 model in a global standard deployment (USA East).

Question

We created a GPT-4o deployment. The Deployed Model version version shows 2024-11-20 but the only available model version via the API are 2024-08-01-preview and prior. Passing '2024-11-20' as the model version returns a 404.

Answer 1

Hi Greg Coyle

I am able to replicate 404 errors with 2024-11-20 version

but succeeds with 2024-10-21 Api version as mentioned in API-version document.

Here was my sample code which is giving outputs

import os  
import base64
from openai import AzureOpenAI  

endpoint = os.getenv("ENDPOINT_URL", "<endpointurl>")  
deployment = os.getenv("DEPLOYMENT_NAME", "gpt-4o")  
subscription_key = os.getenv("AZURE_OPENAI_API_KEY", "<apikey>")  

# Initialize Azure OpenAI Service client with key-based authentication    
client = AzureOpenAI(  
    azure_endpoint=endpoint,  
    api_key=subscription_key,  
    api_version="2024-10-21",
)
    
    
# IMAGE_PATH = "YOUR_IMAGE_PATH"
# encoded_image = base64.b64encode(open(IMAGE_PATH, 'rb').read()).decode('ascii')

#Prepare the chat prompt 
chat_prompt = [
    {
        "role": "system",
        "content": [
            {
                "type": "text",
                "text": "You are an AI assistant that helps people find information."
            }
        ]
    }
] 
    
# Include speech result if speech is enabled  
messages = chat_prompt  
    
# Generate the completion  
completion = client.chat.completions.create(  
    model=deployment,
    messages=messages,
    max_tokens=800,  
    temperature=0.7,  
    top_p=0.95,  
    frequency_penalty=0,  
    presence_penalty=0,
    stop=None,  
    stream=False
)

print(completion.to_json())  

Hope that helps.

Thank you.