HNHacker News
TopNewBestAskShowJobs

arcanemachiner

3,188 karma · joined September 20, 2022

submissionscomments
arcanemachiner··on Kimi K3.1 Pro model weights are out
Too late to edit my previous comment: I meant an uncensored provider.
arcanemachiner··on Kimi K3.1 Pro model weights are out
Side question: Can you (or anyone reading this) recommend a decent source for an LLM provider?

I'll admit that I'm lazy and haven't looked yet... I'm just wondering if you have any recommendations.

arcanemachiner··on The Little Book of Generative AI Foundations: An Intuitive Mathematical Primer
I love how that site offers a phone-formatted PDF as an option. How thoughtful!
arcanemachiner··on Safari introduces Adaptive Dark Mode in 27.3
"Authored"? It just applies a set of rules to the existing stylesheet. It's not like they're customizing it for different websites.
arcanemachiner··on Safari introduces Adaptive Dark Mode in 27.3
There's a mode like this in an Android Chrome fork called Cromite that works surprisingly well. Most of the other options do suck though.
arcanemachiner··on Ask HN: Which TV do you recommend for privacy?
So your idiot brother doesn't connect your spy TV to the WiFi when he comes to visit, instead of using the goddamn Chromecast that is already plugged into it. >:(
arcanemachiner··on AI delivers 'zero financial payoff' to most companies, Apollo economist says
Jealous. I'm getting nerd sniped by silly bullshit more often than ever before thanks to AI.
arcanemachiner··on ChatGPT and Codex usage limits merging
Ah, good point.
arcanemachiner··on ChatGPT and Codex usage limits merging
I can't blame them, I'm a huge fan of Opus 5.5 too.
arcanemachiner··on ChatGPT and Codex usage limits merging
They give (gave?) out resets like candy, I get the impression they pride themselves on how much compute they can give away.
arcanemachiner··on ChatGPT and Codex usage limits merging
Nice, you helped patch the loophole faster.
arcanemachiner··on ChatGPT and Codex usage limits merging
Agreed, ChatGPT is better than Claude chat. Can Claude chat even search the web?
arcanemachiner··on ChatGPT and Codex usage limits merging
I modded my status line to show my remaining 5-hour and weekly limits, and when they reset. It's all available in the request data that comes back with every message.

I don't use Claude chat, but this is a great way to ensure my usage doesn't disappear on me.

arcanemachiner··on ChatGPT and Codex usage limits merging
Oh nice. How many tokens do I get with a Plus subscription per 5-hour quota, or per week? Does it change over time?
arcanemachiner··on Anthropic blocks users for anything touching training larger models
Not familiar with scare quotes, I take it?

And yes, I am taking full advantage of the scammy subscription scam, because it's the best value by far until the inevitable rugpull. Guess I'm not as smart as you and your "grownup" setup though.

arcanemachiner··on Anthropic blocks users for anything touching training larger models
Bad decisions compound over time, so I want my model making as few of them as possible. Opus 5.5 is the current GOAT, and it's fairly affordable on a subscription. (I wish it was cheaper, but nothing else delivers comparable results for my price/time investment.)
arcanemachiner··on Anthropic blocks users for anything touching training larger models
You mean 3.1 Pro now, I assume? K3 is nice, but more expensive than an Claude subscription on a token-for-token basis (until the US frontier rugpull eventually comes), and not as good for my needs unless I'm trying to do "naughty" things, like reverse engineering.
arcanemachiner··on Anthropic blocks users for anything touching training larger models
There is nothing open-weight that even comes close IMO. The only half-decent model that 99% of people can actually dream of running locally is Qwen 3.8 27B.
arcanemachiner··on Anthropic blocks users for anything touching training larger models
I think Kimi K3.1 steals the crown from K3, no?
arcanemachiner··on Anthropic blocks users for anything touching training larger models
Oh, I see it. But there is nothing that competes with Opus 5.5 for 90+% of the work I do, so I use Opus 5.5 for that 90%.
arcanemachiner··on TCP Use-After-Free: CVE-2026-98432 / CVSS 9.8
Out of memory?! That's terrible!
arcanemachiner··on Show HN: Ghostman – A fast, keyboard-driven TUI API client and webhook inspector
Devil's advocate, it could have been a bunch of squashed commits. I've done that when open-sourcing stuff from work, after making sure the code didn't have any stray references to work-related items.
arcanemachiner··on Show HN: Ghostman – A fast, keyboard-driven TUI API client and webhook inspector
1) Sometimes you need to look at things yourself. You do verify the agent's work for important things, I hope? You've never seen them do mind-numbingly stupid things that warrant a bit scrutiny from time to time?

2) To me, Vim provides a normal UI. No slow, bloated, clunky, mousing, no slow Electron garbage, blegh. Just a quick, clean, lean UI.

I can relate to the part of your post where you have difference of opinion, but the apparent solipsism of not being able to understand why someone would want something different than you is the part of your post that drives me nuts. There are different people who like different things, is that not fairly straightforward to understand?

arcanemachiner··on Meta censored the No Kings Instagram account during a nationwide campaign
That and "They trust me. Dumb fucks."
arcanemachiner··on Rubber Duck I've Ever Had
More like title mangling.

You can edit your own submission titles BTW. (Hint to the submitter, if that window hasn't expired.)

arcanemachiner··on Decompiling Fast Food Tycoon 2: Drawing the World
Not OP, but I have to agree with them. The text just makes too much sense. There's no irrelevant shit tossed in for no reason, and I'm not really seeing any annoying LLM-isms.

Of course, I might be wrong.

arcanemachiner··on Artificial Analysis Intelligence Index v5
Devil's advocate: Their scoring system is itself a meta-benchmark.

What happens when you release a new benchmark? Models magically start scoring better against it.

How do you overcome this? Release a new benchmark and start the process over again!

All other things being equal, they should retarget their prices if models start gaming the process to top their index.

Has this happened? Maybe. Is their new scoring system biased towards the American frontier models? Quite possibly.

I have my suspicions, but they have led me astray before. I will continue to be cautiously cynical, though.

arcanemachiner··on Tailscale SSH: Investigation and remediation update
This is true, but I still wouldn't discount the possibility of mistakes. It's good to know how your home network works. Even Opus 5.5 can make some pretty dumb mistakes, particularly if it doesn't have the full picture.
arcanemachiner··on Websites Should Be Printable
This might be more of an edge case than people who complain about websites that require JS, which is itself a number that rounds to basically zero.

Maybe you could capture the full height of the website as a screenshot? There are extensions that do this, and there is some degree of native support for the feature nowadays.

arcanemachiner··on No signs of AI payoff outside the tech sector
> BI

Business intelligence

← PreviousPage 2 of 34Next →