🔎 Should ChatGPT tell us which model actually answered?

ChatGPT
, , ,
1

I think ChatGPT should show not only which model the user selected, but also which model actually generated the answer.

For normal everyday use this may not matter much.

But once people start comparing models, testing reasoning quality, reporting bugs, doing research, or trying to reproduce results, it becomes important.

For example, if I select one model, I would like to know whether:

  • that exact model answered;
  • the request was routed to another model;
  • a fallback model was used;
  • the reasoning mode changed;
  • context was compressed or partially dropped;
  • some other internal routing affected the result.

Something very simple would already be enough:

Requested model: GPT-X
Served model: GPT-X
Reasoning: High
Fallback: No

This information could be hidden under a small “response details” button so it would not clutter the normal interface.

I think this would be especially useful for:

  • model comparisons;
  • bug reports;
  • scientific or technical work;
  • reproducibility;
  • understanding unexpected changes in answer quality.

Right now, when a response behaves very differently from what I expected, I often cannot tell whether the model itself behaved differently or whether something changed in routing, context handling, or fallback behavior.

I am not asking for internal implementation details.

Just a clear answer to a simple question:

Which model actually produced this response?

Would this be useful to anyone else?

2

Hi @evgeny.monakhov,

Thanks for sharing this! A small response details button showing the model used, reasoning level and any fallback sounds useful, especially for comparing models and reproducing results. Keeping those details tucked away would also make them easy to check without cluttering the chat.

We’re taking this as feedback for clearer information about each response. I appreciate you explaining what you’d like to see and how you’d use it.

-Sandra