Wrong OpenAI model used?

[bolidesoft]

Hello!

I have the "OPENAI_MODEL=gpt-4o" setting in the .env file. However, I noticed that replies are not that smart as in ChatGPT GUI. 
My doubts increased when I asked my bot, "What language model are you using, and what is the latest date you are aware of?". 
The reply was, "I am based on the GPT-3.5 model, and my training data goes up to October 2023."
At the same time, I see "gpt-4o-2024-08-06" in the usage stats for my API key.

Is it normal?