Google announced that the new pricing goes into effect by the mid-September. At the same time the Python 2.7 runtime is not ready, and we still don't know what concurrency levels will be allowed (neither does the development team). There's still no word about guaranteed CPU resources per instance (overselling is a possibility). There's no way to prioritize requests from users over task queue requests either.
The old-new billing comparison in the admin panel is horrible as well -- no numbers for new billing in the overview. No way to see how close you're to the new free quota limits either.
The end result is that I'm not even worried that I will have to pay too much when the new billing kicks in, but that I'm completely in the dark about how the things will change after the change and no idea how well the new pricing will work at scale. Supposedly the old one was bad for estimating costs, but the new one is way, way worse at the moment.
This is a complete failure to make transition to the new model smooth. I'm looking forward to the someone explaining in a couple years what kind mismanagement caused all this nonsense.
-Sergey