Hacker Newsnew | past | comments | ask | show | jobs | submit | BrokenCogs's commentslogin

Clone the repo and ask your llm of choice to audit any security risks

After that I run them in containers, no network access, ro filesystem.

Not very practical for a solution that requires a browser plugin. The browser is where most of my interesting data is.

And then build from source. And make sure it doesn't auto-update so no new malware gets in. This is my idea of a lot of extra work.

Building from source is nowadays writing one sentence in a chat box. Your idea of extra work is either unbelievably lazy or perhaps outdated by a few years.

Why even ask the question?

good luck auditing all transitive dependencies

Gemini 3.8 flash seems to (subjectively) be the outlier in terms of performance to cost ratio?

Gemini 3.8 Flash results are often not very coherent but it does put a lot of shading and details to hide the fact.

Its giraffe / grandfather clock one is pretty bad... (two necks? wearing a suit?)

Weird as well, it's clearly pulled out some 1884 patent on clock designs, and a quick ddg/google doesn't show it as anything to do with grandfather clocks.


Google Gemini are the only team who have openly had staff deliberately spend time on SVG performance: https://twitter.com/sunjiao123sun_/status/202455551655137292...

See also this Jeff Dean tweet showing off their animals-in-vehicles abilities: https://twitter.com/JeffDean/status/2024525132266688757


It’s very likely they all add svg generation into the training data. It’s part of the reason it’s no longer a good benchmark (unless you need to generate SVGs).

It's about more than just training data. Every model since GPT-2 had SVGs in the training data, because they all used a scrape of the Web and the Web is full of SVGs. What's unique about Gemini is that they actively worked to get better results for their SVGs - maybe RLHF, maybe RLVF of some sort.

Hugged to death

Yeah, the blog post itself loads for me but everything else is really struggling.

If only ...

Somewhat of an aside but, personally, I also found GTA V and Red dead redemption 2 to be boring because these games didn't respect my time. I could play for two hours and still feel like I accomplished nothing except for "interacting with the world".


'I could play for two hours and still feel like I accomplished nothing except for "interacting with the world"'

Funny how this is exactly the same sentiment I feel about NMS. Go figure, different strokes, ya know?


RDR2 feels to me like a game that would be awesome to play once I'm in retirement and have all the time in the world.


> I could play for two hours and still feel like I accomplished nothing except for "interacting with the world".

Two hours is like three or twelve story missions, depending on where in the story you are. Did you spend most of that time farting around doing stuff that wasn't main story [0], or are you just not really into the stories being told? IME, Rockstar's 3D games have always been not-all-that-great third-person shooters, but made up for it with the storytelling and world-building.

> ...because these games didn't respect my time.

If I were Global Overlord for a month, I'd permanently banish the sentiment "This video game doesn't respect my time". It's fine to say "This game bored me." or "I just didn't enjoy this game.". Unless we're talking about predatory "free" to "play" games [1] that survive by deliberately reaching deep into the wallets of whales, mixing pride into the discussion and claiming the injury of disrespect is very silly.

[0] ...of which Red Dead 2 has an absolute assload -most of which will bore folks to tears who don't really dig the "late 1800's outlaw going on about his day" thing-, and GTA V has a fair bit...

[1] Games that you have to pay up front to play can also belong in this category.


RDR2's intro absolutely did not respect my time. I sat down with an hour free to play it, and in that time, I did not reach a save point. There was some story, but it was very slow, and I had precisely zero interest in repeating it. I ran out of time, turned it off and never came back.


Same. I really want to play it but - I feel I need to skip the first hour to do so?


Almost all of the story missions are: 1) walk/drive/ride to this place 2) Talk to NPC 3) Do an extremely boring task (follow someone, drive somewhere, etc) 4) Walk somewhere else 5) Talk to someone else. 6) Collect XP or another reward.

Most of it is not enjoyable gameplay.


Yeah. Like, the train heist and other cool gameplay segments were fun. But the missions to go to some random town and talk to someone and escort them back? Horrible.


Don’t forget the top notch drawer rummaging simulator.


They're not the first. Each lab has its own: grok bot, codex desktop, perplexity computer, Claude whatever


Only Grokbot, openclAw, hermes, and instinct compare right now due to the mobile-first form factor.


Yeah. Codex and Cowork are full of papercuts when it comes to usability. Grok Bot is head and shoulders better and more pleasant to use. And Muse is similar to Grok Bot.

Claude Whatever would actually be a pretty good product name


Interesting that all of the bikes are turquoise colored, except for the medium effort


GPT-6 is so good that all pelicans born after today will look exactly the one generated by simonw


It's clearly an extension of the previous naming: Luna, terra, sol



Content not viewable in your region.


:(


For a uv noob like me: what do you mean by bundling it all into a single binary? How does that work?


UV can be built (or downloaded) as a single optionally statically linked with the libc executable file, with no other dependencies. That makes it very easy to distribute on weird systems.


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

Search: