Getting back on topic....
First drive on 10.69.25. Some good, some bad.
On TX-66, the car passed up the opportunity to incorrectly shift into a left turn only lane mid-block that has been a reliable regression under 10.69.3.1. And, it correctly made the turn from TX-66 onto the Bush Turnpike Frontage Rd then managed to shift two lanes left and enter the turnpike with no interventions. Over the last year, every time this has been fixed, the car manages to find a new way to fail here. So, it passed both of these problem areas once in a row. Not good enough to call fixed, but promising.
However, if the car is in the left lane of a multi-lane highway (not NOA), the car still likes to jump into a left turn only lane that opens up as you approach an intersection. This has been curious behavior since the car intends to continue straight and there is no reason for the car to conclude that it is compelled to use it. After all, if the car is now in a middle lane, it should be able to assume that the lane will continue straight through the intersection. Neural networks are fun...
Had to use accelerator a couple time to hurry the car up to make a merge out of an ending lane. When there is a car in the lane we want, but too close to us, the car gets stuck - especially if its moving at close to the max speed. Tesla needs to add an exemption to the max speed that permits the car to briefly exceed it a little bit to get enough space for a lane change. That's what humans do and what other cars expect.
I had a couple interactions with other cars on residential unlaned roads that went fine. I didn't feel that the car was too far left and was not uncomfortable.
I did have some unexpected slowdown for an overhead blinking yellow light on a highway using FSDb. The previous version would slow very slightly for this light, but not enough to be a concern or warrant intervention. This time I had to use the accelerator to override the braking action.
NOA on a high-speed 2-lane road was unchanged from previous versions. Worked fine except for annoying wandering within the lane. At no time did the car attempt to cross the lane divider, nor did I feel like it might.
I played a bit with the new auto-cancel turn signals. They worked as intended, but I only had a couple uses as I ride on FSDb almost all the time where they would be useful.
I do have one gripe about the GUI. As it sits today, the ADAS indicators are scattered even further around the display than before. The blue wheel is at top left, the TACC max speed is near the top center and the NOA indicator is now at the very bottom center. I think that Tesla needs to put some thought into how to unify these indicators into the same location on the display. I don't know the solution, but there has to be a better way than scattering them around. Related items should be grouped together.