When will EAP w/ HW2.5 stop getting better

[SeminoleFSU]

I'm still seeing improvements but I'm wondering we'll get these code improvements where Elon says the system is moving from 2D to "4D" (e.g. 3-dimension tracking + time)

I'm hopeful we'll see EAP continue to get smoother/better but we just won't have the surface street goodies. What do you folks think?

 

[Tam]

I'm still seeing improvements but I'm wondering we'll get these code improvements where Elon says the system is moving from 2D to "4D" (e.g. 3-dimension tracking + time)

I'm hopeful we'll see EAP continue to get smoother/better but we just won't have the surface street goodies. What do you folks think?

It seems like the upcoming FSD release is a different module (city) so that should not affect the already existed highway module.

As the highway module gets better, I expect that would benefit both FSD and EAP up to a point.

That's because EAP is designed with a human driver while FSD not.

Up to some point, FSD including on highway will surpass EAP.

 

[bdub85]

They will have to support HW2.5 and older as long as they exist in the fleet. I would think of it more as a parallel effort. They supposedly have 300+ engineers working on AP, so I think they should have the bandwidth to do both up until they can push HW3+ only releases. This is all just my wild speculation though.

 

[SeminoleFSU]

Yea my guess (or better yet hope) is it's one code base with different parameters saying which chipset and features are active. In a HW2.5 car you could have no AP, AP, or EAP. My hope is that each time a new build is pushed, any logic, code refactoring, etc improvements that apply to the feature we have will improve even though we're on HW2.5
Obviously at some point the HW2.5 will be forked and abandoned (like HW1 was) but my hope is that will be a good while and not upgrading to FSD in my car won't be something I regret. As long as EAP continues to improve (which it has) I think I can live with that...

Any new car I buy (considering the tri-motor CT) will have FSD... but I may trade this one in on that so I don't think it's worth it to upgrade to FSD in a car I may not have for long...... Wish Tesla offered a way to have FSD go with the owner, not the car.. :-(

These price increases feel a little like a mafia tactic to be honest.

 

[bdub85]

These price increases feel a little like a mafia tactic to be honest.

It's 100% a shady profit grab. I don't see how you can justify raising the price for something that just went to beta testing. I have no issue charging more for a new feature, but if I pay 10k vs 8k after Thursday I get no additional features. The whole FSD pricing just rubs me the wrong way. Hopefully the market will push back.

 

[shipapotamus]

It's 100% a shady profit grab. I don't see how you can justify raising the price for something that just went to beta testing. I have no issue charging more for a new feature, but if I pay 10k vs 8k after Thursday I get no additional features. The whole FSD pricing just rubs me the wrong way. Hopefully the market will push back.

Yep, I was planning on upgrading to FSD at some point at 8k before increases, as I just got the car and wanted to slowly check things out. These fomo tactics don’t work well on me, it feels similar to a dealer telling me that x car is hot and they are worried it will sell before I decide. It ends up shutting me down mentally to purchase.

 

[bdub85]

Yep, I was planning on upgrading to FSD at some point at 8k before increases, as I just got the car and wanted to slowly check things out. These fomo tactics don’t work well on me, it feels similar to a dealer telling me that x car is hot and they are worried it will sell before I decide. It ends up shutting me down mentally to purchase.

They released this beta and definitely tried to FOMO people into buying it before they up the price. Don't fall for it. I am still pissed I wasted 6500 buying FSD on my first Tesla. I didn't recoup that money on trade in either, so it's not an appreciating asset as they claim. I didn't know any better at the time.

 

[SeminoleFSU]

It's 100% a shady profit grab. I don't see how you can justify raising the price for something that just went to beta testing. I have no issue charging more for a new feature, but if I pay 10k vs 8k after Thursday I get no additional features. The whole FSD pricing just rubs me the wrong way. Hopefully the market will push back.

100% agree. They may not even get gov approval.. I could wreck my car between now and the hardware upgrade, or between now and the software rollout... It's shady as hell.

One other thing that is confusing to me is I went back and looked at the spec sheet and what I paid for my car originally. I paid $5k for EAP. When I go into the mobile app under upgrades, it says FSD is $5k for me... So i'd be paying $10k for FSD anyway.... I guess that means if I wait until after the FOMO bully squeeze TOMORROW I'd be paying $12k all in? Nah. Think I'll pass.. hence this post- how much more improvements do you guys think we'll see with HW2.5 and EAP?

 

[SeminoleFSU]

Yep, I was planning on upgrading to FSD at some point at 8k before increases, as I just got the car and wanted to slowly check things out. These fomo tactics don’t work well on me, it feels similar to a dealer telling me that x car is hot and they are worried it will sell before I decide. It ends up shutting me down mentally to purchase.

Amen. I don't like feeling bullied. When I do, I walk away. Did that on a house we were looking at this weekend.

 

[ken830]

I'm kind of on the fence with this for my 2018 Performance 3. I think I'm going to pass because my early 2013 S is approaching the 10-year mark to be replaced in a year or two and will definitely have FSD.

But I was always half expecting Tesla would be forced to upgrade the HW2.5 + Enhanced Autopilot cars anyway -- I mean, our EAP has the summon feature, which I assume will have to use the 4D NN to safely navigate a busy parking lot, right? Unless Tesla just assumes our summon will be subpar compared to HW3 summon... But hmm.. they'd probably need to give it two separate names to distinguish what feature the customer actually has.

 

[Chanceofthat]

I'm kind of on the fence with this for my 2018 Performance 3. I think I'm going to pass because my early 2013 S is approaching the 10-year mark to be replaced in a year or two and will definitely have FSD.



But I was always half expecting Tesla would be forced to upgrade the HW2.5 + Enhanced Autopilot cars anyway -- I mean, our EAP has the summon feature, which I assume will have to use the 4D NN to safely navigate a busy parking lot, right? Unless Tesla just assumes our summon will be subpar compared to HW3 summon... But hmm.. they'd probably need to give it two separate names to distinguish what feature the customer actually has.

I was in a similar position, 2018, EAP on HW 2.5, and I finally decided to purchase HW 3.0 because after reading though various opinions on here and Reddit, there seemed to be a large number of people who thought that Tesla would consider EAP/ Existing Smart Summon “feature complete” and not provide any further improvement.

None of us has a crystal ball so we won’t know until it happens. Did I just waste 5k? Maybe. Did I prevent myself from spending 7k later for something I could have gotten for 5k? Also maybe. Sorry, I just realized none of this is helpful . In the end I decided that there were too many unknowns and that given I was financially able, I might as well.

 

[SeminoleFSU]

I'm kind of on the fence with this for my 2018 Performance 3. I think I'm going to pass because my early 2013 S is approaching the 10-year mark to be replaced in a year or two and will definitely have FSD.

But I was always half expecting Tesla would be forced to upgrade the HW2.5 + Enhanced Autopilot cars anyway -- I mean, our EAP has the summon feature, which I assume will have to use the 4D NN to safely navigate a busy parking lot, right? Unless Tesla just assumes our summon will be subpar compared to HW3 summon... But hmm.. they'd probably need to give it two separate names to distinguish what feature the customer actually has.

I wouldn't hold my breath on Tesla upgrading HW2.5 compute to get promised features to work. See HW1 owners and promised features

 

[ken830]

I was in a similar position, 2018, EAP on HW 2.5, and I finally decided to purchase HW 3.0 because after reading though various opinions on here and Reddit, there seemed to be a large number of people who thought that Tesla would consider EAP/ Existing Smart Summon “feature complete” and not provide any further improvement.

None of us has a crystal ball so we won’t know until it happens. Did I just waste 5k? Maybe. Did I prevent myself from spending 7k later for something I could have gotten for 5k? Also maybe. Sorry, I just realized none of this is helpful . In the end I decided that there were too many unknowns and that given I was financially able, I might as well.

I also just bought it just now. Last year, I was really close to buying it for $2k... At the time, I was thinking that even with the promised HW3 upgrade alone, it was worth $2k if it meant EAP features improved... even if I never saw FSD during my vehicle ownership period... But it's hard to make that decision when HW3.0 wasn't even out yet, and I kind of just let the deal go without pulling the trigger.. Of course, I can justify tonight's purchase as a deal because I'm always buying TSLA since 2013, and I like to think that $2k from last March went into TSLA and is now worth over $16K... Even taking out the $3k difference and capital gains taxes, I still come out way ahead...

I wouldn't hold my breath on Tesla upgrading HW2.5 compute to get promised features to work. See HW1 owners and promised features

Agreed... Now, I guess I don't have to since I bit the bullet and just bought it.

 

[Fernand]

I don't see how the "FSD" label has any meaning, since autonomy is a gradual development that will never stand still long enough to be called "full". What we've seen so far w/ Tesla is that there's Freeway NOA and now Street Level NOA, and these are different code branches, optimized for different issues.

I suspect that HW2.5 will see continued improvement to the Freeway NOA functions, because it reduces Tesla's liability exposure. Because of limited computing power, it will never run Street NOA code, just "lane keeping AP" and an improved Summon. But why wouldn't it improve?

Completely autonomous unsupervised Robotaxis will probably evolve as something different from a "car". A "car" would remain a vehicle that you can take anywhere on or off "the grid", freeways, roads and city streets, in the woods, on unpaved roads etc with generalized driver assistance AI optimized for independence. Cool. That's what our Teslas are.

Urban Robotaxis will probably operate in constant communication with each other and a central. Safer and more efficient. Maybe have roll-on roll-off capability with a longer range mass transit system. Maybe operate in their own lanes. Or using tunnels.

It might be possible to repurpose a "car" as a "robotaxi", but it would probably involve hardware changes way beyond HW3 vs. HW2.5.
.

 

[powaking]

I also just bought it just now. Last year, I was really close to buying it for $2k... At the time, I was thinking that even with the promised HW3 upgrade alone, it was worth $2k if it meant EAP features improved... even if I never saw FSD during my vehicle ownership period... But it's hard to make that decision when HW3.0 wasn't even out yet, and I kind of just let the deal go without pulling the trigger.. Of course, I can justify tonight's purchase as a deal because I'm always buying TSLA since 2013, and I like to think that $2k from last March went into TSLA and is now worth over $16K... Even taking out the $3k difference and capital gains taxes, I still come out way ahead...

I'm still kicking myself for not jumping in when it was priced at $2k. I'm not entirely sold on FSD nor do I see myself using it all that much, what I was really interested in was the HW3 and having my existing EAP experience improve. With the mentioning of FSD subscription maybe just subbing for a month or 2 will force them to install HW3 and that would be my cheap way into getting the new hardware but not actually dumping $7k now for FSD.

 

[FloridaJohn]

I suspect that HW2.5 will see continued improvement to the Freeway NOA functions, because it reduces Tesla's liability exposure. Because of limited computing power, it will never run Street NOA code, just "lane keeping AP" and an improved Summon. But why wouldn't it improve?

What happens if they just merge (or more accurately, expand) the Street NOA to also work on the freeways? I suspect this is the ultimate goal because it adds so much functionality that would be useful on a highway. For example, being able to detect and move around stopped vehicles. I can also see that happen with the Summon functions as well.

At that point, you now have all three major parts of the "self driving" software (highway, city streets, and summon) running on HW3 code. At that point I think we will start to see a ramp down of new features for the HW2.5 folks. Their cars will continue to do what they do today, but will no longer get the new updates and features. This is similar to what is happening to the AP1 folks now.

 

[Fernand]

@FloridaJohn good points. Who knows, really. It depends how the code was structured, and written. The modularity. All we can do is guess at this point. It could be that the Freeway code works with higher speeds, and that at 85 MPH even HW3 can't quite do all the Street processing to the sides that's key to intersections but isn't needed in Freeway mode anyway. Maybe that's the boundary?

 

[SeminoleFSU]

I'm still kicking myself for not jumping in when it was priced at $2k. I'm not entirely sold on FSD nor do I see myself using it all that much, what I was really interested in was the HW3 and having my existing EAP experience improve. With the mentioning of FSD subscription maybe just subbing for a month or 2 will force them to install HW3 and that would be my cheap way into getting the new hardware but not actually dumping $7k now for FSD.

I thought about this but from everything I've heard the subscription FSD is intended as a way for users to try it out for a brief period to see if they think it is worth ponying up for the full amount. If that's the case, I highly doubt Tesla is going to come upgrade the hardware so someone can try it for a brief period. From everything I've heard, this is a multi-man hour process by a mobile tech and then there's the cost of the hardware itself. So i'm skeptical that would make financial sense for them. If they did offer that, I would jump on it just to get smoother EAP if I didn't feel like FSD was worth whatever the number is next week...

 

[SeminoleFSU]

What happens if they just merge (or more accurately, expand) the Street NOA to also work on the freeways? I suspect this is the ultimate goal because it adds so much functionality that would be useful on a highway. For example, being able to detect and move around stopped vehicles. I can also see that happen with the Summon functions as well.

At that point, you now have all three major parts of the "self driving" software (highway, city streets, and summon) running on HW3 code. At that point I think we will start to see a ramp down of new features for the HW2.5 folks. Their cars will continue to do what they do today, but will no longer get the new updates and features. This is similar to what is happening to the AP1 folks now.

I agree- that makes sense John.

 

[Chanceofthat]

@FloridaJohn good points. Who knows, really. It depends how the code was structured, and written. The modularity. All we can do is guess at this point. It could be that the Freeway code works with higher speeds, and that at 85 MPH even HW3 can't quite do all the Street processing to the sides that's key to intersections but isn't needed in Freeway mode anyway. Maybe that's the boundary?

Karpathy talks about the structure here a little bit @ 18:40