written-beyond

I don't know what curse is on my openai account. Since ChatGPT 4o their desktop app does that weird extreme lag slow down in almost any sized chat other than completely new, like 1-2 messages deep. I changed some settings, the problem got better and then immediately broke after they released the codex app(new chatgpt app). It is literally completely unusable on Windows, apparently works fine on mac for my friend. Never had this issue with anthropics desktop app.

I find it ridiculous how lean is treated like some magic pixi dust by people now.

I hope you find relief from your disease.

I got our pocket god ChatGPT to fact check this discussion. No lab is profitable, it is a moonshot to expect they'll ever recover the billions frontier labs have put in behind these DCs. I never understand this argument everyone tries to make, AI is too expensive -> It will get cheaper as they scale -> AI is too cheap and not sustainable -> No it's very profitable on API -> go back to 1.

I am not an AI denier, I use it and I find it very useful. That said I can disagree with the current economics of the industry and their business model.

Define profit? If by profit you mean covering compute costs then yes you're correct. If by profit you mean covering spent RnD budget and future RnD budget allocations then no, by no means are any of these AI companies running at a profit.

Also these third-party providers are running magnitude smaller models than what Anthropic and OpenAI's power users expect from them. Yes the average chat message isn't being routed to the 1+ Trillion param model but people running CC or Codex on Ultra probably are. Those are objectively more expensive.

It doesn't on windows FF

This is the first time, in my entire life I have experienced a website overriding pinch to zoom. I couldn't pinch to zoom, just on top of pictures, which would expand the image then zoom into it.

Why would the author do that, very frustrating.

They're all building the "Her" companion device. It's almost like those transparent touchscreen phones you'd see in SciFi movies that every phone maker would have their research Labs working towards, Nokia morph was one of them.

Those things all look this in concept and in movies, but their practical utility in real life always seems to be lacking. I'd imagine this is the reality what everyone building companion devices are facing.

With the amount of money, RnD and engineers these firms have this seems to be the only reason.

I mean cartesia did one of their first demos on a Mac book pro, live.

Still extremely impressive but not unheard of.

Too late I just shipped tsvhdl so you have a hardware implementation of the actual TS compiler. Why steal valuable CPU cycles from agent workflows, off load all transpilation and type checking to a dedicated TSC chip.

lol, I made the same comment the last time this website was shared here.

After reading the title, but before visiting the link, thought this was a backup of all of Terrance Taos work.

You're right but also very wrong.

You can never prove that there aren't any bugs. The definition can also change according to time and expectations.

I've made tens of internal developer tools because I couldn't gaf about using some third party vendor, all before vibe coding. Even then we purchased software from vendors, and we still do it now.

You can't cherry pick 2-3 pieces of problematic software, imply that it's how everything works. You're vibe coded Photoshop will never be used seriously, unfortunately. Same goes for the video editor, if you're fine with an autonomous AI agents listening to some internal github repo for issues that get auto or human reported and attempting to provide live fixes and feature enrichment, you aren't going to get that.

The problem really does stem from the fact that these agents, no matter how good you keep making them are still limited by a time horizon. AI companies have gotten really good at extending that horizon. The contexts get longer, prompts get less specific but theres a deadline to all of this. Take that LLM point it to a specific problem let it run, pull it out start new. The LLM will inevitably accumulate technical debt in it's own context, that starts to create a rot. You can try to improve this problem by making sub agents, etc. but you will 100% run into context rot after a period of time for the lead agent.

They aren't, they did some much needed API changes after adopting signals. It reduced some of the magic but it's pretty much the same as it's always been. I agree that API stability is something a lot of mature frameworks need, but svelte has always been at the forefront of provided high quality migration tooling.

We weren't asking which tools integrated LLMs, we specifically asked which got better by building them with LLMs.

Cursor, Codex and Claude Code all have terrible desktop applications, which are a laggy mess on Windows.

What improvements did UV do with auto research loops, I wasn't aware of that.

I thought this was posted on a Dbus hate website when I saw the URL.

Night and day, thank you!

Not doubting you, care to share what those improvements were?

can you update this website, I just wish the entire layout wouldn't shift when the page gets loaded and the timestamps in the title look very ugly and take up a lot of space.

Insane, I've never used those versions of macos before.

Sometimes I think I'm Morty speaking with the death crystal but I don't even have a death crystal and all I fear is life itself.

I hadn't read the article and read this comment as though NVIDIA themselves were implying that this library was checked but not trusted by them since it was fully LLM generated.

I want to, I have like a 2000+ day streak on Duolingo and I know I can only gain fluency once I start speaking it more.

I thought that was on purpose...

Those models can never be capable enough to compare to the level of output generated by frontier labs. Where would you even host these models? Maybe a company like Shopify could manage self managing an few instances but that's it's own headache. Unless Shopify can show significant productivity gains by using ChatGPT or Anthropic's API pricing models there is no way they'll be able to justify using those services at those prices in the long term while maintaining their employee pool.

The other option is to just stay with ChatGPT and Anthropic and fire almost all of their employees, but then they're running the risk of having a significantly worse customer experience in case your development cycle gets out of hand. They'd have lost most of their internal knowledge in the layoffs so they'd be stuck.

If they don't do anything and just let their employees continue to work like they've always been but with LLMs to enhance productivity. Not to the point of implementing software factories, but as most people are using them in the industry. They lose nothing, gain productivity but their shareholders, investors and managers have been gaslit into believing if they're not 200x-ing their AI spending they're probably going to be left behind by competitors who are doing that.

Also some of their larger investors may have investments in other AI companies and they probably strong arm them into contracts with those companies to benefit their portfolio.

Absolute garbage article and garbage website. This content shouldn't be allowed on HN.

Svelte did always have a runtime, they're still doing this from what I understand the major part of Svelte 5 was moving to signals.

HOW DOES IT WORK FOR YOU PEOPLE?!

Istg on windows, I can't get codex app, new chatgpt or chatgpt classic to work without hanging on even a moderately sized chat. I tried all of the different tricks people recommended but that shit just crawls. It was stable at some point, classic that is, but then it started being a laggy mess.

I've uninstalled new chatgpt/codex app. I can't access phone chats on it, I can't use the desktop allocated chat messages on it. I have to use up codex limits. I am so surprised that OpenAI ships that hot garbage on windows.

I guess you could say, you weren't cimpress-ed

Secret Hitler(the boardgame) fan?