I use Claude-4-sonnet’s thinking model. After 4-5 conversations, it switches from this model to the normal model. Who knows, maybe I’m not even using Claude-4-sonnet. Even if the chat bar Screenshot by Lightshot says I use it, it doesn’t actually use it.
when I open a new chat I can see that it is using the thinking model. you can see this when it warns that it is already thinking. But as the conversation goes on, the AI stops using it.