Just watch it, 52secs in, first crash... So you know, before you think its a 100% fix. It might help at lower speeds (the new motors make it go a little bit faster)

Many thanks for posting the video.
I tried to make a hypothesis about the tumble problem, different from what others say.
I think it is not due to a structural or design problem and therefore from an unstable condition due to the passage of air, such as Vortex ring state, but I think the problem is due to some problem on the IMU and magnetometer.
The IMU calculates the heading by seeing the velocity vector how is displaced in angle and compares it with the heading seen by the magnetometer.
If for some reason there is a bug or any problem, even magnetic interference, you can run into problems similar to the toilet bowl or, when very sudden, to an overturning if the two measures do not coincide.

We'll see what DJI will do.

Thanks for sharing.  Definitely not the fix many were looking for.

Thing is, it's a cinewhoop at heart. Somewhere along the line since this style of drone and traditional kwad were released, people have lost sight as to how they are designed to fly. They aren't freestyle drones, the idea is for smooth and steady. All these motors do is provide a greater powered motor to enable it to go as fast as a traditional fpv kwad and do the acro moves it just isn't designed to do. Not to mention voiding any warranty that DJI have when you purchase the drone. Yes, there seems to be an issue with some drones with the spin of death or other issues when performing some moves, but was it designed to fly like this ? I will wager the answer is no.