heaney-555

The benchmark you linked to shows GPT-6 Astra having the lowest hallucination rate of all tested models.

The DC-X was incredible for its time, but it failed.

> but if you are talking about something reusable that takes off and lands vertically

No, I am talking about something that takes off and lands vertically after boosting a first stage to orbit.

Suborbital is 100x easier.

Having the idea is not the hard part! Making it work is.

Tesla Autopilot hasn't been updated since 2019, it's just a lane-centering feature. You're probably thinking of FSD Supervised, which is indeed Level 2.

But Tesla Cybercab, taking paid rides in Austin right now, is Level 4. And has no LiDAR. You can fly to Austin and take a ride on it right now if you don't believe me!

The Model 3 was the first mass market long range EV.

The Nissan Leaf models before the Tesla Model 3 had a maximum range of 107 miles.

The Tesla Model 3 had a range of between 220 to 310 miles, depending on the trim.

This is what made the Tesla Model 3 revolutionary. It was the first to be mass market with long range.

The Model 3 was the first mass market long range EV.

The Nissan Leaf models before the Tesla Model 3 had a maximum range of 107 miles.

The Tesla Model 3 had a range of between 220 to 310 miles, depending on the trim.

This is what made the Tesla Model 3 revolutionary. It was the first to be mass market with long range.

Fine, let's be more clear: reusable rocket boosters that land themselves.

Experts said it wouldn't work. Then they said okay maybe it works, but the economics won't. Now they say nothing.

You don't think the human race will be extended to Mars?

You don't think the human race will be extended to Mars?

As for frontier AI model: let's see Grok 4.8 before drawing any conclusions there.

SpaceX's FFC application is for 1 million Starmind satellites.

>It’s also not happening.

People said this about every one of Musk's big ideas, from Falcon 9 landings to Model 3 mass production, Starlink, and FSD.

You should look into the Artemis program! Not only are we going back, but we're building a base, which will effectively be the spiritual successor to the ISS.

"Behind" is doing a lot of work in this headline.

The Chinese models are benchmaxxed to hell. They aren't as good in practice as all the hype suggests.

These agent swarms are from inside OpenAI, with the safeguards built into the public API disabled.

Russia does not have access to this, and as with all western tech companies, AI providers do what they can to prevent Russian usage of their products at all.

As for open-source models, Russia's electricity grid is under severe strain with the Ukraine war, and only recently has it started building out serious sovereign compute capacity.

A 2.5 MB download just isn't a problem in 2026. I would far rather see the lossless image than save a few MB.

Shorter robot vacuums with similar capabilities are arriving on the market relatively soon.

This is the classic fallacy of imagining a future problem with current technology.

By the time we have "one-million-person cities" on the moon, synthesizing water or hauling it from icy asteroids will not be a problem.

>rather than bullshit like “self driving”

You should go try it. Do a free test drive with FSD next time you have a free hour. You'll stop using the scare quotes.

>and having ad companies surveil my intimate space

So don't buy one from an ad company then? There are home robots that run entirely on-device.

Your Roborock is not the state of the art.

https://www.wired.com/review/matic-robot-vacuum/

I would argue that as of 2026, we have enough storage and bandwidth that lossless compression is justifiable in almost every use case.

Computers should not be making images look worse than they were captured. It's not 1996 anymore. We don't need to be doing this at all.

Is this going to be the cope every time this happens?

Again, within the project sure. But there is fierce competition between different open-source projects!

Tesla charges a $10/month subscription for its connected features, with relatively high uptake because its software is actually great (unlike the dogshit UX of most cars), so it doesn't need to sell data. Plus it just isn't how Tesla operates.

They are. These aren't Flock cameras. The post is a lie, but it's one people want to believe.

Is everyone going to ignore that those are not Flock cameras?

Why is the quality of discussion on this topic on HN so poor? I'm not even pro-Flock, but a blatantly false post like this is just pathetic discourse.

The article has no mention of Flock

They objectively did not. You should go and read the report!

Sure, but you still wouldn't describe the student as just doing what they were told to do.