Comment by Imnimo

Just that it's another model where you can read the raw "thinking" tokens, and they sometimes fall into this sort of rut (as opposed to OpenAI's models, for which summarized thinking may be hiding some of this behavior).