I clearly didn't turn on the auto-select function,The model is not the one I used, but the official website shows the one I used

Today when I was using Claude3.7thinking, I found that the reply was wrong, so I asked about its model, and it told me that it was Claude3.5. And I didn’t turn on the auto-select function. Even though I used slow request, I wanted to use Claude3.7thinking, even if I had to wait a little longer. But you secretly modified the model! Please give me an explanation!!! My request ID is b12f065f-afb7-447f-8a10-2cd762f8ddeb!!!I’m a pro user and now I’m a little worried if this is also the case with the quick requests I used before.

1 Like

@danperks

Hi, thanks for the question.

The model doesn’t actually know what model it is! The reason it replies with Claude 3.5 sonnet is likely due to that model being the last known model within its trained memory. Claude 3.7 as a concept didn’t exist before the model was trained!

This doesn’t happen on the Anthropic website, for example, because they specifically tell the model its name in its system prompt.

I guarantee that whenever you pick a model specifically from the list, that is always the model that is replying to you!