ctolsen
ctolsen on Hacker News
Sagrada Família is an amazing building. Going inside it is an awesome experience.
But it is a _lot_. Putting aside the artistic skill and construction cost of it, I'm glad I don't live in a city where everything is like that. I feel like it'd be overstimulating much like an analog version of the Las Vegas strip.
They do have shareholders, but the majority is held by the Proton Foundation.
I've seen the show many times through and I know which episode every single one of these are from. But with a few exceptions I don't remember the name of any of them. It's just not a thing my brain picks up.
> Are there modern harnesses that don't do subagents/subthreads/whatever the same way?
At the end of the day it's a UX question. I could do the same as this (I use Paseo, similar thing, less obviously vibed copy in the docs if that helps...) with a bunch of terminal tabs and pi, and maybe some skills and scripts to help out, but it's a pain to manage.
Thanks, signed up and found it.
AI is not the issue here. But when your business is based on security and privacy you better adopt it carefully and deliberately. If you're measuring your teams on how much AI they use, I don't trust that you're putting security front and center. AI can be used to improve security, but you can absolutely do the opposite if you push too hard.
Also, any time anything comes out it gives a clear impression that the leadership are bullies, from the CEO on down. That creates an environment in which people are afraid to speak up, which means issues get overlooked. Which isn't great for security either.
I would have absolutely no problem with a CEO setting a solid, well-founded direction about the use of AI to improve their products while keeping a foundation of security and privacy. But Yen just sounds like a Silicon Valley tokenmaxxing douchebag.
I run my business on Proton and we do miss some things. But they’ve been doing a good job making things happen slowly but surely, and I can be patient.
What I can’t tolerate is if they stop putting security first, which it sounds like they are under Yen’s leadership.
Is that "specific kind of API key" something that can be used in any agent, or is it Vibe CLI only?
The thing that makes the least sense is that Haiku 5.5 is apparently better than every single open weight model in existence except Kimi K3.1. Which is just... not true. I feel like I don't even have to explain it to anyone who has touched both sides of that equation for more than three minutes.
Opus 5.5 is even more verbose. 260M on the same suite.
I mean you can run them in the cloud and not worry about safeguards.
Why do you need the open weight models to work at home? You can use them whenever you want.
I am the opposite. I work with LLMs all the time but don’t use them much at all personally. Not for lack of trying, I’ve tried several times with different frameworks and I just can’t find a use case that is remotely worth it.
I've used OpenBSD the last year or so (as a bastion/jump host for a few systems) after using *BSD systems very little for probably the past decade, and... good lord I love it so much.
Say what you want about the benefits of systemd or the usefulness of other parts of what is now unavoidable with Linux, and I won't argue with many of them, but OpenBSD just feels like a *nix system is supposed to. `ps aux` is like 10 entries at first. `pf` makes so much sense. `man` pages are consistently incredible. The security/cryptography focus, managing services, pledge/unveil... so many things about it is just *chef's kiss*.
It's not and will probably never be a daily driver for me and it may not fit your use case for anything, but I highly recommend playing with it.
Should do it across the board. The world would be a better place if harvesting personal data for ad targeting was illegal.
I’m currently running Paseo for agent orchestration for code, beads managing tasks, and OpenWebUI for stuff that’s more on the business side (with RAG or MCP to our drive, CRM, etc). It’s mostly human triggered though it doesn’t have to be. Paperclip might be a more LLM-managed option.
All of it is self hosted, models are Ollama Cloud and OpenRouter. You can throw dirt cheap models at a lot of these tasks and they perform very well. It’d be no effort at all to replace inference providers.
Every time any detail comes out of Anthropic or OpenAI it usually turns out whatever they’re doing is just incompetence. Right now I’d put money on that anything they were doing with git was incompetent garbage and they decided that the path of least resistance was to vibecode this thing.
Can't wait until this botnet working at 0.005 tokens per second kills humanity in a billion years
> The almost 744B parameter count model capabilities will "in the future" fit in 8B parameters. That's their assumption?
The article says by the end of the year. It's laughable.
This entire article is underpants gnomes logic, but somehow media will pick it up anyway.
To be fair to them, on this particular benchmark (Harvey's Legal Agent) 20% puts it almost top of the pack of _any_ model, and certainly top for any remotely in its price class.
Opus 5.5 scores 4% on it.
Not sure "freaking out" is the word I would use, but it’s fairly obvious looking at OpenRouter usage that the price cuts on Luna a while back were in response to intense competition from dsv4.
So the industry is responding, where it matters. Which is on heavy API usage, not coding subs.
I’ve been using it a lot and it’s performing really well. Not GLM 5.3 levels but it beats Deepseek for my use. I’ve used it on long running coding tasks, though mostly prototyping, but it’s done a great job at very low cost.
ADHD is highly treatable with medication, and any other intervention is at best mild mitigation. So how many? Amongst people with ADHD, basically nobody.
Sun and proper sleep is certainly helpful for people with ADHD but the likelihood of them getting that is vastly increased if they’re medicated correctly.
400 million is pocket change. For all intents and purposes the US is no longer supporting Ukraine.
GLM 5.3 is very impressive and definitely better, but it also at least 4x the price.
On that note I’ve been subbing in MiMo-2.6-pro when cost is an issue, which is super cheap and also performing really well.
I find this stuff so annoying. The way you deal with these people -- abusive snowflakes -- is that you say no. We've appeased and appeased and we get nothing in return, Europe is carrying the entire financial burden for supporting Ukraine (which was a reason to try to keep the US happy) and we just get more tariffs every time Trump has a tantrum.
When people step over the line, tell them to fuck off. This is true in personal relationships and international relations alike. Sometimes there is more nuance but in situations like this it's clear that nothing else works.
European countries detain and deport immigrants who are illegally present all the time. The level of it can be discussed, but what's happening in the US is that they've detained hundreds of citizens and deported immigrants who are present legally because they said something the people in charge didn't like. I'll skip that part, thanks.