Today's Claude 3.7 seems to be a bit abnormal

The thinking model never thinks. The output is very brief and nothing like when I used it yesterday. Often it fails multiple times on a problem.
When I switch to Anthropic Console, the model solves the problem in one go.

seem to be just the 3.7 or 3.5

Why is 3.7 thinking when it not showing how it think? Is it really 3.7 thinking but the thinking is skipped as it already thinks in the background as it develops? We need answers