Azure OpenAI API model inaccurate?

I am using azure openai api with gpt-4 (and 32k).
When I ask it what model it’s running I get this back:
“I am powered by OpenAI’s GPT-3 model, not GPT-4. As of now, a GPT-4 model is not available. The specific capabilities may depend on how OpenAI has implemented and updated its system.”

However the response model metadata states it’s running gpt-4.

Hi @aleldowf - gpt-4 (incl gpt-4-32k) is trained with data up until Sep 2021 when gpt-4 was not yet available/released. Hence, it does not know about its existence.

The metadata is always the most reliable source.

The exact model that is being used for chat completion is returned in the model attribute of the chat completion response object

Whatever the model claims is just a hallucination.

if you want to give the model a persona, you can pass that within a system message.

e.g. “You are gpt-5 an AGI” and the model will assume that persona