Hacker Timesnew | past | comments | ask | show | jobs | submit | daemonologist's commentslogin

Pricing: https://dev.meta.ai/docs/pricing-rate-limits

Interesting that they have separate API pricing for "we can train on your data" (whereas iirc most of the big players either make that distinction only between subscriptions and API usage, or train on everything). Wonder how it compares to Deepseek V4 Flash given that they're similar on pricing and data policy.



One admirable part of Waymo's service area in Dallas is it serves areas previously devalued by racist redlining. This is in contrast to Uber's Avride and Tesla's Robotaxi offerings, which stick only to the richer, whiter areas of Dallas https://techcrunch.com/2025/12/03/uber-and-avride-launch-rob... https://dallasinnovates.com/tesla-launches-robotaxis-in-dall...

I doubt the geofence exclusion is because of the demographics of those areas and not the demand for autonomous vehicles in those areas, although I don't doubt the latter is a result of the former.

I doubt this is correct. “It’s expensive to be poor” — a lot of people who can’t afford cars (or have bad credit) end up Ubering to/from work every day

How can that be anywhere near financially sustainable.

If you are curious about this, I recommend There’s No Place for Us: Working and Homeless in America. It covers how some folks, with no credit, are forced to live in extended stay hotels (where they take cash up front and don’t run credit) which can cost multiple times what rent can cost.

If you put the author, Brian Goldstone, into YouTube, you can find some recent news coverage of some of the folks from the book and iirc one of them is Ubering in a lemon, working 60-80 hours a week just to break even.

In America, unfortunately, we’ve perfected the debt trap and turned poverty into a profit center. The same equity companies own rental properties, extended stay hotels, and storage locations so they can vertically integrate homelessness.


The most evil thing I heard was companies like Doordash and Uber buying up data on who has credit card debt, so they know which driver is more desperate for cash, so they can low-ball them and give them a worse offer because they're more likely to take it because they're desperate for it.

Where is the proof of any of this? There are too many conspiracy theories about gig work. There was a similar post on HN a few months back[1] based on some anonymous vagueposter on Reddit, again without evidence. Would I be surprised if this is true? Of course not. Should we just start believing random things without any critical thinking? No.

^1: https://qht.co/item?id=46461563



You fell for some kid's story and should feel bad for repeating it unsubstantiated.

Wow that's terrible but also pretty smart...

>How can that be anywhere near financially sustainable.

It's not and it's indeed a very sad statement on public transport. My father has spent some time in his retirement Uber-ing and said it was shocking to him how many repeat customers he got that were having to spend so much just to get to a minimum wage job.


It's not. It's expensive to be poor.


How can that be anywhere near financially sustainable.

All you have to do is basic math.

Monthly car payment (which is higher if you're poor) + monthly gas spent (which is higher if you buy it in a poor neighborhood) + monthly insurance (which is likely to be higher if you're poor, and/or live in a poor neighborhood) + maintenance is very often far less than what it costs to take a rideshare ride 14 times a week.

When I drove for Uber, about 20% of my rides were people commuting to and from work who could not afford a car.


Aren't you going 20 times per month, twice, that's 40 rides.

At an average 25$ you're spending 1'000$/month on Uber.


Aren't you going 20 times per week, twice, that's 40 rides.

Work 5 days a week = 10 rides. Plus four rides on the weekend, or not. That's why I used 14.

Where do you get 40 rides per week?

At an average 25$ you're spending 1'000$/month on Uber.

Even if someone did 40 rides per week at your posited average of $25/ride, that's $4,000, not $1,000. Are you letting an LLM do math for you?


They showed their work and named their units. Their math is fine.

If work happens 5 days per week, each workday requires two rides, and there are 4 weeks in a month, then there are 40 rides per month. If each of those rides costs $25, then:

5 × 2 × 4 × $25 = $1,000, per month


But he didn't state per month. He stated per week, then went back and changed his comment after I replied.

Guess you didn't notice that I quoted him.


You're so right! That's such a great load-bearing detail!

I especially liked the part where you quoted "1'000$/month" and got yourself all confused about the intended conclusion anyway!

Good chat, my friend!


40 rides a month, not week. I'm not sure you're in a position to accuse folks of not doing math when you don't seem to read very well.

Read what I quoted. He edited his comment after I replied.

You knew very well there was a typo in the first sentence, and decided to look foolish none the less.

It was clearly a per month calculation.


Carpooling?

This is somewhat similar to the situation here in LA where they service and very often drive through Westlake/MacArthur-Park which undoubtedly low income. I think it’s for practical reasons though, as they’re not allowed to use the freeways here yet. So for people going from rich areas (Bel-Air/Beverly Hills) to DTLA they need to cross/drive through there anyways.

Even so, I’ve encountered plenty of people who live in those areas and have been able to take Waymos which is great given how sketchy that area can be…


Ish. The problem with Waymos is there's no human driver, so they'll adjust pick up and drop off in a way that could be dangerous for the passenger. A human driver isn't going let a passenger off into a dangerous situation. They'll block traffic to drop the passenger off at their destination instead of a few blocks away in a dangerous neighborhood.

Waymo's got the legal go ahead to use freeways, but they're working on some of the subtleties of self-driving on the freeway vs city streets before enabling it for everybody. (Specifically, California car culture is you merge left early instead of staying in the right lane until you have to. Apparently the Waymo driver was staying in the right lane.)


> The problem with Waymos is there's no human driver, so they'll adjust pick up and drop off in a way that could be dangerous for the passenger. A human driver isn't going let a passenger off into a dangerous situation.

In this situation, you absolutely can (and are encouraged to) press the support button in the car to talk to a human agent. They have the ability to tell the car to go somewhere it normally wouldn't go for any situation that humans can reason about.


Tesla has fewer miles than I have driven in my lifetime. It’s currently still a half baked product. This might be a story in a few years but today you’re comparing an F150 to a Delorean.

How many miles is that?

The linked qwencloud page has pricing; it's $2/6.

https://www.qwencloud.com/models/qwen3.8-max


Thanks, totally missed that!

It seems to only be available in Kimi Code, via subscription, no there's no API pricing. The linked page says it consumes about half as much quota as the 1M version though.

It's not available via API (yet).

Btw if you do need a permissively licensed model for this task, there are lots out there, e.g.:

https://github.com/KupynOrest/s3od

https://github.com/Tennine2077/PDFNet

https://github.com/PramaLLC/BEN


Agreed, and even SSR always felt like cheating imo. I wanna see someone build a primary radar on an esp 32.

Gotta admit that's what I clicked through hoping for.

I've seen a few projects that use SDR to detect doppler shifts of reflections of broadcast FM radio station off planes as "passive radar". But that doesn't get much in the way of direction info, just approach velocity. I've pondered building something that uses that and correlates it with an ADS-B receiver, to see it it can detect aircraft that are not transmitting ADS-B. But that'd get me on another "list" if I published it I suspect...

This is not the project I remember, but it's the same idea and technique: https://github.com/Max-Manning/passiveRadar


I suspect it's possible to build an active radar without exceeding the limits of the ISM bands.

Actually, I know it's possible, that's how the automotive radars (used for adaptive cruise control) work. The real question is if someone can get enough range to pick up aircraft.


Is it bistatic or monostatic? How do you know that it is doppler and not multipath?

My read is that they're announcing a plan to build non-folding wings to test the geometry they would eventually want with folding wings. Which makes sense but isn't very exciting.


It's not just geometry, it's also where the mass is. And the rigid areas.

Aircraft wings oscillate during flight, the location (and movement) of mass within them (e.g. fuel) is critical to model.

Also, the location of the fold is likely less flexible (e.g. more rigid) than the remainder of the wing. This will greatly affect the dihedral angle of the remaining outboard portion of the wing.


It might be that a lot of the cost doesn't show up in the data because it comes out of your paycheck before it's "yours," in the form of your employer's contribution to health insurance premiums. (Anecdotally this is true for me - "my" portion of premiums would not be in the top three, but the total premiums are #2 after housing (for a while when I had roommates they were #1, which is ridiculous).)

(BLS gets this data from surveys: https://www.bls.gov/opub/hom/cex/home.htm )


Correct. You can argue this is money that you’d otherwise get if it wasn’t going to healthcare, and that’s a fair point. But then you have to count that in the top line income.


Parakeet is way faster (on Nvidia hardware) but not quite as accurate in my experience.


It's also super fast on CPU.


There was a fad a while back of building insanely long prompts - tens of thousands of tokens - including having models write prompts for themselves. I always thought it was counterproductive, especially if you're going to use the prompt more than a couple of times. (That said, the e.g. Claude Code system prompt is insanely long, so if you genuinely have a lot of information to provide maybe it's beneficial. Like, shorter is better, but you don't want to be under-specified.)


Consider applying for YC's Fall 2026 batch! Applications are open till July 27.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: