Addressing the GPT-3.5 to GPT-4 Switching Issue

When using the GPT-4 model and reaching the specified token limit, the conversation automatically switches to GPT-3.5. However, when returning to the original conversation after the time limit, the dialogue model does not automatically switch back to GPT-4. How should this issue be resolved? I believe this is a counterintuitive setting, and OpenAI should fix this bug.