ehsankia

LLMs should be for general/high level work. For example, an LLM with the right harness could fetch and use that CNN, or train it's own CNN or other model to use for this task.

The harness in my experience makes a huge different too. Out of curiosity I tried having Gemini chat (with Argon) try to solve a level of this new puzzle game I'm linking (Order of the Sinking Star). It just got in an infinite loop and crashed. Then I try Antigravity with Flash 3.8 and it solved it easily. I saw the harness looked at other sokoban puzzles, coded some stuff, tried to replicate the level, etc. All things that a pure Gemini Chat doesn't really do.

LLMs are great for the high level harness, but if you want to count letters for example, it will still rely on "tools" like a piece of code to do it more reliably. Similarly in the above example, you wouldn't have the LLM directly try to analyze the images, but instead fetch and use the best tool for the job. But that can only be as good as the harness/tools it has access to.

It's indeed very confusing, it's either

1. It's available to Pro in Gemini app, but NOT antigravity (for coding)

2. It's a phased rollout and Ultra got it first, Pro is coming later

The latter blog post is specifically "Argon in Antigravity" versus the other blog post is about Argon in general, so seems like #1 is more likely.

Yeah this is confusing, the announcement pages claim it's both, but the app itself claims it's ultra only... Maybe a slow rollout?

I'm a bit confused, all the announcements say that this is availabe to Ultra and Pro subscribers, but in antigravity, it's grayed out and says it's only available to Ultra.

This was actually started federally under Biden, but thrown out as soon as Trump came in

https://www.ftc.gov/news-events/news/press-releases/2024/10/...

So many other similar examples, like Bank Overdraft Fee Caps, Credit card late fees, personal data safeguards, and so much more.

Right, the better argument would've been to say that we should wait until the drug undergoes some early trials, rather than just "some expert looking at it".

It's exact same price as Sol 6.1 announced yesterday.

I guess if one of them hits singularity, it could in theory just wipe out all the rest, seeing how they keep escaping and hacking into other systems :)

> and google just started letting all their engineers use claude

That's misleading.

1. Having different models available is useful for A/B testing and helping improve Gemini itself.

2. They have an enterprise offering for Antigravity (their agentic coding platform), and they need to test that it works well with non-Gemini models too.

> late last year/early this year

That's an eternity when it comes to coding models.

In my personal experience, we've had almost a step change every ~3 months this year, at least for bigger one-shot tasks. For example looking at Gemini Flash 3.0 vs 3.5 vs 3.8, it went 5% -> 30% -> 75% on DeepSWE, all since the start of the year.

Right, in terms of hardware and price, I've always been tempted by the Meta Quest, but the ecosystem and the fact that it's locked down has always been the major blocker for me. It does suck that this more than twice as much (although some of it is obviously the current surge pricing), but for hardware that I own, it's well worth it.

Curious if there's any info about how much Foveated rendering helps performance roughly. Is it like 50% more frames?

From my experience, it's quite hit and miss for every app.

If you use Google 99% of the time and only check other apps when Google is wrong, then that's a biased experiments.

I disagree.

Similar to you, 95% of the time I do the simple use case that I've memorized. The problem is that even after a decade of using Git, when I get into those 5% where something goes wrong, I have no idea how to actually solve it. None of it makes any sense in my head, and I have to search for a bunch of magical commands to run, or nuke things to get back to a good state.

With JJ, it just makes sense. It's intuitive and clear. I have been able to recover without any help. Just the undo alone is basically a life saver. Rebasing and merge conflict resolution is so much simpler.

Yes I probably won't use it to its full potential, but still, the little I do use makes 10x more sense than Git ever did.

I don't understand why the fact that it's an Airbnb really matters here? Unless a large number of apartments are meant to be empty, why does it matter if the tenant is short term or long term? Is it that short term users leave and enter their apartment more often?

I just tried it, and the answer is non-sense.

I asked it something simple, list some good indie puzzle games, and half the answers are games that don't exist. Imo quality > speed.

Isn't another issue that most successful open models are distilled from closed models, but closed models are putting more and better safeguards against distillation?

No? The only different between these robots and a washing machine is that in theory, the robot is generalist and can do many tasks, whereas a washing or laundry machine can only do one task. You can still in theory fully control what task the robot works on. Also, in theory, the robot would be the glue between all the other machines, like filling up the washing machine, then moving it to the dryer, etc. It deciding what to do isn't a "prerequisite" for the robot.

SpaceX has had 165 launches in 2025 (although admittedly 75% of those were for Starlink...) Obviously bigger isn't more economical or sensical, and most cases are served just fine with the Falcons, but there are cases we need the big boy for, and it's good that someone is working on it and has made so much progress.

Obviously a semblance of a schedule is good to have, but realistically, that's not really how research works. Look at James Webb telescope, it was originally scheduled for 2007, and ended up launching 14 years later. It's still an amazing piece of engineer/science, and it's amazing that it's up there now, even if it was very late to it. It's much better to be late and successful than early and failing.

Even better, since everything is well organized, you can add `opacity: 0.7` to `.wall` specifically, and get something that looks almost exactly like how old school wallhacks looked like.

> Isn’t this what AGI is by design?

Well, the "G" in AGI is kinda important. These are specifically games/puzzles.

> they have to be retrained from scratch

Is that true? Didn't DeepMind already build plenty of agents that are generally good at most computer games without being retrained?

If it's just a tool for myself, and I use a webkit based browser, why would I use a hacky solution for a browser I don't use?

Sometimes I'm developing an internal tool or something only for myself / handful of people. I'm perfectly fine saving time and complexity using a one liner modern CSS solution instead of having to rely on some hacky unreadable code to support 10 years of legacy browsers.

Yep, Tile I believe is the only third party service that exists. All other trackers either plug into "Apple Find My" or "Android Find My Device" network. There's finally starting to be a few devices that can do both, but they're rare, so make sure you get the right one when buying. But they take 10s to setup and it's very smooth.

Atuvos one at 1.6mm and UGreen one at 1.7mm are great, though one time battery is annoying. There are some that have wireless charging, though thicker.

> relatively affordable

You can buy 4 third-party trackers for the price of 1 official one.

They do lack UWB, though there are other great form factors such as cards, and cool features such as wireless charging or usb-c charging, which imo is nicer than swapping batteries every few months.

That's awesome. I'm glad that trackers have reached a price point, reliability and form factors that I can easily put one in everything I care about. I even have card ones in my wallet, my steam deck / e-reader case, etc.

Also, most of these have usb-c / wireless charging, so I don't have to mess with random cell batteries every 6 months.

> So they're like, buying into a depreciating asset

Part of the issue is that the average age of the House is ~55 and for the senate it's above 60. So they have a lot less incentive to care about that, or about climate change.

Anything before L4 is "driver assist", which means at the end of the day, the buck stops at the driver. Anything beyond L4, the car itself drives without requiring supervision, which makes a big difference. It's your responsibility to use lane assist in a reasonable way, it's not your responsibility to control how an L4 drives anymore. That's the point of self-driving, the "self" is responsible.

You could try passing it 10-20 front pages across a much wider time range.

You can use: https://news.ycombinator.com/front?day=2025-12-04 to get the frontpage on a given date.