I paid $5K for EAP back in 2018 when I bought my car...it was more than I really thought it was worth, but I figured I'd roll it into the cost of the vehicle, and I really wanted the ability to be able to have the car change lanes automatically (on command, not NoA type automatically). This was really only useful to me on occasional road trips, so at the time a subscription that I could have enabled for one or two months out of the year would have made much more sense.
But what if I had to start all over today?
Well for one thing, for the level of functionality that I am really looking for, I think Tesla is going to be forced to offer for a lot less than $6K. That level of functionality is available (or will be soon) for much less from the competition. Tesla is at high risk of losing their reputation as a technology leader if they don't treat EAP-like functionality as an included feature (or at least very nominally priced).
But if forced to either pay a one-time fee or a subscription, I can tell you that knowing what I know today, I wouldn't go with either. There are just too many bugs in the system that haven't been fixed in the 4 years that I've owned my car, and no doubt they go back further than that. If Tesla cannot fix these basic issues in that amount of time, my faith that they ever will be able to is dwindling.
The two key bugs (and two functionality misses) are:
- Phantom braking. I will say that I seem to have a lot less problem with PB than most. I just did an 1800 mile road trip and had less than 5 occurrences of PB on the entire trip. And 1 or 2 of them might even have been excusable, as I was coming over a "ridge" that was obscuring the road ahead, and the car slowed down. But even 3 or 4 hard braking events is too many and erodes trust in the system. These have to be once in 5-10,000 mile type events.
- Errant lane centering on unmarked on-ramps. Some states (Virginia!) don't mark lane separators to the end of the merge lane, resulting in what appears to the car as an extremely wide lane that the car attempts to center itself in. Extremely annoying and even dangerous at times. My wife's ID4 does not behave like this. It essentially maintains its position along the left of the travel lane.
- Extremely sluggish acceleration when passing. Sometimes when you are in the slow lane stuck behind traffic, you may need to pull into the fast lane and accelerate quickly back up to speed to merge into the fast lane, but the car takes FOREVER to accelerate. I'm not looking for a roller-coaster like acceleration here, but at least something more along the lines of what a real human would do.
- Lack of acceleration as/after a car departs the current lane. The scenario is that you're stuck in the right lane behind a car that is slowing to exit. The car head is 90% out of the lane and the car continues to slow until the vehicle ahead is completely out of the lane ahead, well past what a human driver would wait before accelerating. I kind of get this one. At one point, the AI system only worked on one snapshot of video at a time, so based on that, it may not be clear whether the car is exit or entering, so it plays it cautious. But how long has it been since Tesla told us that they were incorporating time-based neural network processing that would be able to determine whether the car was moving into or out of the lane?
There may be one additional bug as well, but frankly I've gotten so sick of it I pretty much leave NoA off, and that is the propensity of NoA to pull into the passing lane at off-ramps.
These may not seem like huge things, but on a road trip, it really limits the usefulness (increases the annoyance level!) of EAP. Until they fix those, I'm afraid I don't think it's even worth the $99/month you propose. More like $25-30.