derac

Maybe we should also burn books and lobotomize ourselves to make science more fun and challenging. The ego here sickens me. I don't care about your ego. I want accelerated material science, medical science, energy, better outcomes for people across the world.

I disagree, though I highly dislike the affect of Anthropic models.

This was reduced recently to 2/12. It's also next to impossible to allocate the free stuff unless you put in a CC. If you do, you get the servers right away, though. Just be careful not to spend. You can set up a budget trigger to shut down the vms.

Old servers may have been grandfathered, but mine was set to 1/1 and couldn't be reshaped.

I made this to do cost/performance analysis on the top SOTA models. Maybe you'll find it useful as well.

I am taking all the measurements from the GPT-5.6 announcement blog and presenting them in a tabular format with relative scale against a chosen model/effort.

The json is on the page, it's all static, so feel free to take the data and do what you want with it.

Only? Seems remarkable to me.

Thanks for this explanation!

Btw, I recently did my first patches with ghidra. Is there any way to avoid jumping into a code cave by shifting all the instructions up then updating any statements referring to those memory locations? Seems like it should be simple enoigh in theory, but I didn't find an implementation on a cursory search. Could probably code up a plugin pretty easily.

This is cool, maybe "target left arm" implies that it's your left, when targetting them and you're facing each other. So, their right arm.

I switched from Windows a couple of months ago. I am loving it. I cloned my favorite Windows window management solution, WindowGrid to KDE, https://store.kde.org/p/2363952/

I haven't found anything on Linux like it.

the unfathomable reason is that it is significantly more energy efficient.

Largely it is legal, but it's complicated. https://www.eff.org/issues/coders/reverse-engineering-faq

https://youtu.be/FHufjrP0xDI?is=xmFQrXGa1dBHM67I

This is a helpful explanation of what this technology is and looks like. (Munro)

I think running them against each other with a rules engine would be more interesting. Count up illegal moves and wins/unfinished games. I think llm grading is too unreliable.

Is there some consumer protection law around this?

Look at the table of supported modalities. It can take in input of image/video/text/actions and output image/video/text/actions.

It is unlimited under the free NVIDIA Developer Program. You're talking about a different sort of acct I think. The dev program acct is 40 rpm unlimited for personal use.

I recommend nvidia nim for completely free dev access for young people.

I wasn't trying to be condescending. I was stating my experience talking to people 10-20 years ago, before AI was thought about at all by people outside of tech. It always boiled down to a dualist vs monist ontology argument. I agree there are valid reasons to dislike current generative AI tech though. I agree with a lot of this.

I should say I do think that deeper ontological thing is why people tend to think the tech will always be a novelty or will stagnate soon, etc.

It's funny, I have the opposite experience of everyone around me hating AI. I'm not aggressively pro-AI around them at all but you aren't allowed to have any positive or nuanced opinion of the tech.

I'm used to it though, I've been excited by the concept of AI since reading about Turing and such as a child 20 years ago. The idea has always been met with negativity, IMO because people want to feel that they have a part of themselves that is beyond nature and has a "special" place in the universe.

According to Google Wikipedia still gets 4 billion pageviews a month. The article seems a bit hyperbolic. There are certainly concerns around the nature of work and the economy, though. There are of course ongoing concerns about global warming. I'm not denying that, but I don't think it's particular to AI tech.

Why can't you use technology and communicate with your friends? The same can be said for how people abuse TV, fast food, etc. It's up to you to live with care and attention or not.

On the other hand, tech in general (not just AI) does make an easier and easier path for people to go inward and neglect their community/family/friends. This does suck.

Why is it Rube Goldbergesque? The process doesn't seem arbitrary.

I think Honesty can be evaluated. Does the model push back when it knows the user is wrong? How often does the model hallucinate data vs. say it doesn't know? Provide a prompt with contradictions or other issues and see if the model corrects you.

Here is an article by Anthropic that explains what they do and mean in more detail: https://alignment.anthropic.com/2025/honesty-elicitation/

Yes, say something. It's worth the awkwardness to potentially save someone's life. Just approach it delicately.

When 3 came out they mentioned that flash included many improvements that didn't make it into pro (via an hn comment). I imagine this release includes those.

I haven't run it, but I looked through the repo. It looks very well thought out, the UI is nice. I appreciate the ethos behind the local/offline design. Cheers.

This is faster if their marketing is right, it uses significantly less tokens. Gemini 3 flash is very good as well.

I agree, they seem to be following the Apple playbook. Make a closed off platform and present yourself as morally superior.

Any chance Voxtral Mini Transcribe 2 will ever be an open model?

Kimi k2.5 is a good choice.

I really like the agent swarm thing, is it possible to use that functionality with OpenCode or is that a Kimi CLI specific thing? Does the agent need to be aware of the capability?