The GPT-4o is using GPT-4 as a model

Since the last update of GPT-4o, ChatGPT displays the new model, but the answers are still generated using the GPT-4 model (as shown in the image below). Additionally, the voice feature remains basic, only transforming voice to text and text to voice, without any emotional expression.

1 Like

Hello! I have the same problem. Have you ever received a meaningful answer to this question from somewhere? How long do we have to wait for the real 4o

It is GPT-4o, but it is not trained on the fact that it is 4o as it’s training data doesn’t exceed when it first came out, as that wouldn’t make since.

I dont understand this, sorry

LLMs (Large Language Models) do not have real-time knowledge.

For example, let’s say that GPT-4o was released in February, but its training data went up to January. This means that it wouldn’t have any knowledge of its own name, as those blog articles, etc., that talk about it haven’t been written yet. This means that because all data regarding GPT-4o was released later than February, it wouldn’t have any knowledge of them.