sparse-Matrix

I love the application. It became my default driver overnight. But the zig dependency and the odd installation paradigm make it a complete bugger to upgrade.

FYI I'm using debian trixie, fully updated, and feel like installing/updating this should be a no-brainer.

I cam here to say, "Let me guess: It's all the AI bubble"

What about the other servers?

404 comments indeed. The https://release.files.ghostty.org link given to obtain buildable tarballs is '404 not found'.

absolutely fascinating.

Why am I not surprised that Heinlein wrote this?

The guy was an incurable optimist.

Requiring a membership, free or otherwise, is a goddamn paywall.

Your point is taken, and I suppose I should have been articulate; but these are all huge models, beyond the ability of most of us common folk to load up on a PC and run.

And to be honest, many of those models are not free to use from their providers.

That is, unless you have the hardware to be a provider yourself.

Do that again, and show me all the 37B and smaller LLMs that have made the scene, and I think you'll see what I mean.

Its a far shorter list.

At one time accelerator keys were considered mandatory for a fully realized interface.

Link to a gguf? tensor files? anyone?

TIA

Sorry, this is a hard pass for me. If I need an account on your server to use it, I don't need to use it.

I always seem to be 'that guy'... but what this really amounts to is cloudflare enticing the users to go 'all-in' on cloudflare, drawing them into and integrating them with the cloudflare ecosystem in a way from which they will never be able to extricate themselves.

Customer capture at its finest: "Go with us! its a good thing you'll love it because you'll never get away".

Good stuff.

"oof, written by AI, I be struck plumb dumb doH" - half the HN commentors, probably...

---

Its really simple. Larry went long AF on OpenAI, and OpenAI can never be profitable.

It isn't rocket science; it's first year finance.

It's really bizarre how many people come to Hacker News to bitch about LLMs

This is a common misconception among those who nothing about AI

Honestly, why do you care so much if AI was used?

Assuming he did use AI, it didn't spontaneously generate the project and foist it upon the user and the world; the user used the AI to generate the project at his direction.

It's like the old friend when I decided I didn't care for wearing a full beard in July and got really angry with me about it when I shaved -- not my friend anymore.

Dont like the project? move along.

Otherwise, why are you wasting your energy to put down this hardwork?

All of this can be addressed by using locally hosted models running in bespoke harness rigs.

Bravo! I never would have thought this would be so simple. As a direct consequence of this exercise: I now know something about what, precisely, htmx does.

cheers

I've been having better-than-most performance using qwen3.6 to write python/flask.

I tried using it to generate some rust code yesterday, and it generated much code but only ever came within 1 error of a testable build. The 4th or 5th full rewrite is sitting in the buffer right now.

I'm currently looking to up my game with Bottlecap AI's return of qwen3.6, 'thinking cap'.

Alleged to be twice as fast and superior at coding over extended sessions (vs. 3.6).

We'll soon see.

This may come as a surprise to a lot of AI concerns, but I have -zero- interest in paying for a model.

This puts google in the circular chain of SillyCon Valley businesses under-writing/financing each other's debt.

That said, as indicated by the presence of Fidelity Investments, everybody is in on it at this point.

Unfortunately it crashed out 'no space left on device' while installing the python demo/quickstart.

Only problem was there is plenty of space on the device. PLENTY (not quite 750gb).