mixes stats for 0.12 and 0.13 mess up 0.12 and 0.13 stats.

Looks fine to me. The graph is cumulative: the 0.13 users (yellow) are displayed on top of the 0.12 users (green). However since there are so few users on 0.13 (only a few beta testers), there is very little space between yellow and green, so the green is not visible.

You can see that the same happened with the 0.11 -> 0.12 transition: there was a long period of beta testers using 0.12, where the green is overlayed on the purple, before 0.12 is releases, where the purple drops away sharply. The same will happen on Monday when 0.13 is released, where the green will drop away sharply.

What makes you think that it’s mixing up 0.12 and 0.13 stats?

If it is cummulative, then it explains everything.