I made something that tries to highlight the humor regarding this:

https://news.ycombinator.com/item?id=48950534

Can confirm, just now.

I wanted to make something topical that highlights the humor surrounding the AWS billing issue. Please don't hesitate to let me know what you think.

Incredible work, more effort put into this than the AWS QA Team had

I love Forgejo!

I am somewhere in the middle, where I want something with more than 48GB/$2k of VRAM, but less than 384GB/$40k.

I'm curious if GMKtec's EVO-X2, with ~96GB of usable VRAM, is still a good solution for something like this for $3,399.

I picked up the 128gb version when it was $2,199 and it runs Qwen 3.6 reasonably well with a 128kb context. Not very useful for complex tasks but it can handle some web stuff.

I've been happy with an OEM Spark (128G), enough so that I picked up a second one. Have 2x qwen and 1x gemma (both at 8bit and full context), plus embedding, Re-Ranker, and a 1.7B for little things. Running 6x models, probably going to add STT here soon, want to try talking more than typing.

The caveat is that if you try to use multiple models on the same device at the same time, you thrash and destroy tok/s

It has lower memory bandwidth than most comparable Macs.

I literally said "Oh, of course" out loud.

My average eventually made it to about 3900, and then stagnated between 3600-3900. I'm curious if this is universal behavior or not. I'm up to about 5k steps.

The repository now appears to be disabled.

https://github.com/DO-SAY-GO/ghostbox-releases

https://github.com/crisdosaygo/ghostbox-home-reveal

Yeah folks on this thread mistakenly thought it was against ToS and flagged it to GitHub. I’m confident and trust GitHub will bring it back once they look into it. It’s not a ToS violation.

This idea is great in concept, and I think it's important to state that, but the GitHub Actions stuff is against TOS iirc + they will need to address that pretty quickly.

I appreciate your "great in concept" part.

ghost aligns with ToS: it's a CLI porcelain driving the user’s own GitHub account to use Actions minutes for their own dev workflows.

There was abuse on this thread tho - not just the fake accusations against ghost and me, but people abusing the GH flag/report button on its repos, leading to them being auto-disabled. I trust GH will restore it once they look at it.

In the meantime - if there’s a specific clause you think applies, you're welcome to share. I think it's a classic case of crowd madness.

I believe this has the same issue as the last article that had these claims.

We can assume that Mythos was given a much less pointed prompt/was able to come up with these vulnerabilities without specificity, while smaller models like Opus/GPT 5.4 had to be given a specific area or hints about where the vulnerability lives.

Please correct me if I'm wrong/misunderstanding.

> We can assume that Mythos was given a much less pointed prompt

On what grounds can we assume that? That's what the marketing department wants us to assume, but what makes us even suspect that that's what they did?

Carlini's unprompted talk is one source: https://www.youtube.com/watch?t=204&v=1sd26pWhfmg

>On what grounds can we assume that?

because the bugs they discovered were yet undiscovered?

Or did they hire a team of cybersecurity specialists with the vast amount of funding at their disposal? I don't think its reasonable to assume they used none of their other resources to search for something that could be a very profitable marketing campaign.

They say the focused prompts come from a previous step where the same model "planned" how to discover bugs in said repo. So it might be something like "here's a repo, plan how to find bugs, split work into manageable chunks" -> spawn_agent("prompt" + chunk).