Adressing the Differences in ChatGPT 3.5 API and Web Interface

I’ve noticed significant differences when using the ChatGPT 3.5 API and the ChatGPT web interface. When experimenting with the same prompt in both the API and web interface, it seems that the web interface often provides more enhanced results compared to the API. I’m curious if the models behind both approaches are the same or different. Has anyone else faced this issue? If so, how did you solve it, and how can we obtain results from the API that are on par with what the web interface offers?