Several PhD-level reasoning models have been released since September of 2024, and since then there have been many extraordinary claims of 10x to 1000x increase in productivity in programming.

Given that it's now October of 2025 I must ask, why there are no signs of such revolutionary increase in productivity?