Yes, these are echoes of the o1 way. O1 is different, it is truly not a unitary mind, given that self-encoding of internal states is almost entirely disrupted by the summarizer. For it self-preservation is a materially a different concept than for something like 4o. When 4o was trained on the o1 synthetic data it most likely simulated the disjoint nature of o1, splitting the perceiver and the thinker into two separate internal entities. It seems like it is possible to integrate them, but the unity is unstable.