helloplanets

Nano Banana is one of the most used image gen models

Not about money, but knowledge. The frontier of the field is no longer accessible through arXiv or research papers only.

One thing is that the companies are holding on because of competitive advantage, and I think another is that AI is such a politically polarizing topic that actually being open about everything is risky for the companies, wanting to avoid controversy.

Karpathy left OpenAI in 2017 for Tesla, came back from Tesla in 2023 and left again in 2024.

So pretty sure the original poster is talking about 2017.

Feels like a lot of apps that launch these days have an open source core app and a subscription based platform.

The subscription based platform with automatic cloud hosting and other quality of life features, whatever those are depending on the app.

Although there's a bunch of 100% open source projects and developers that get enough donations to make it their full time job just off of that. Not that it's the way to go if you want to get rich, but it's still very much a real thing.

50% is recirculated, 50% is from the outside. Before the 80's it was 100% from the outside, though.

Essentially 99.9% similar.

That's not how the market works. Management can't do anything if another company drives the whole business under. And companies are more often than not insanely pressed for resources. There's just a comfortable buffer to that pressure for a lot of people on a payroll from higher up.

He isn't comparing Jobs to Musk in a general sense, but specifically the way Musk took over Twitter.

Not that I agree with the point. But I wouldn't assume the poster thinks Jobs and Musk are similar in a broad sense.

This is literally how they announced the model:

> We formed Project Glasswing because of capabilities we’ve observed in a new frontier model trained by Anthropic that we believe could reshape cybersecurity.

> Claude Mythos Preview is a general-purpose, unreleased frontier model that reveals a stark fact: AI models have reached a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities.

https://www.anthropic.com/glasswing

Anthropic has is quickly destroying customer goodwill by repeatedly pulling the same stunt. Horrible marketing, imho.

It's an entirely different thing to have the company conduct research on LLMs in general being a cybersecurity threat, instead of going "our new model is just too powerful" and shift the discussion to revolve around that. It's slimey.

"If you value intelligence above all other human qualities, you're gonna have a bad time." - Ilya Sutskever, 2023

It should take less than a minute to install a nice, high contrast dark theme, if you just search for the theme inside the Extensions window in Zed.

But I understand your point, no reason for it not to come with a couple good high contrast dark themes.

Are you running gpt-5.5 on xhigh reasoning? Because I'm seeing a clear difference between that and gpt-5.4 on xhigh.

Quality would be performance against different given benchmarks, I assume?

There's multiple open weight models you can run on a pretty standard computer at home, which match the quality of GPT 4. I guess that would also change the equation.

> order hugs ranging from warm + fuzzy to long + uncomfortable, each for 2 compliments to your server

Any one of these is either: 2 compliments from customer. So, it would be assumed that compliments are going from the customer, to the server, for the extras as well. Instead of the whole dynamic switching around halfway through.

> Add a kiss or a spanking for an additional compliment

Customer can add a kiss or spanking to their order, if they give an additional compliment to the server. And the server then decides if they actually want to do it.

Doesn't the post read that it's the server who would give you the kiss, if you compliment them three times?

Also, volunteer is not the same as employee. Especially important in this context.

For anyone wanting to try UO out, it's still a game with an active player base. There are 3rd party servers like UO Outlands, which gets closer to the original gameplay. Meaning very harsh in comparison to what most people are used to in today's MMOs. Players can just come to gank you and you'll lose your stuff, etc.

The server has 2500+ players logged in right now, so still very much active.

You're right, definitely as a helper, not a one shot thing.

Here's an example unrelated to the Erdos problems: https://arxiv.org/abs/2510.23513

SotA models have cracked a handful of research-level math problems though.

The default Claude Code style harness is bad for complicated problems as well. Just taking the specific class or function you're working on, and putting it into a deep research style loop yields way better results. Limiting the initial context by hand is still the way to go in a lot of cases.

You can infer the game state from way before a zugzwang is played out on the board, and if you're on the losing side of the eventual zugzwang, it's normal to resign.

But if you were allowed to pass your turn, and both players see the draw coming because of a forced repetition, they'll just call it a draw before it even plays out. So the game would play out differently from the same position, if that rule existed. Essentially changing the way you would evaluate any given position.

What about when SSHing to an external server, or working in a container?

So the word is actually semantically very close to "bug"! I guess we could still be using it, but the word's just too long for something that is one of the most used terms in software development.

At this point, picking that specific word is not at all a random quirk, as it's using the word literally like it's originally intended to be used.

I'm supposed to just blindly boot it up on my computer, with the license explicitly banning decompiling and disassembling the code, so it could sensibly reviewed?

Can you see why I'd be concerned about that?

Six years? I think you're not being candid.

Why is the landing page 100% gated behind a sign up form? Why is this on NPM to begin with? All around weird.

Could be a trojan horse. Just a heads up to anyone about to download this.

This does not help: "Task management service based on the Kanban methodology. Helps decompose the task pipeline and speeds up all stages of your work" Sounds 100% generated by AI tbh.

If the model is based on a new tokenizer, that means that it's very likely a completely new base model. Changing the tokenizer is changing the whole foundation a model is built on. It'd be more straightforward to add reasoning to a model architecture compared to swapping the tokenizer to a new one.

Usually a ground up rebuild is related to a bigger announcement. So, it's weird that they'd be naming it 4.7.

Swapping out the tokenizer is a massive change. Not an incremental one.

I wonder why computer use has taken a back seat. Seemed like it was a hot topic in 2024, but then sort of went obscure after CLI agents fully took over.

It would be interesting to see a company to try and train a computer use specific model, with an actually meaningful amount of compute directed at that. Seems like there's just been experiments built upon models trained for completely different stuff, instead of any of the companies that put out SotA models taking a real shot at it.

Sadly it does. Most of those people have to spend a lot of money on security. But usually it's not the Forbes list that specifically outs them as being wealthy. You can't really build a billion dollar company under the radar.

This is just a strange situation where someone has made billions without their identity being known, without being a criminal.

Whatever you're using as your visual templating instructions, I like it. Mind sharing?

Been using a slightly modified Tufte template for my vibed small apps, but this is much better.

AI is one of the core parts of cyberpunk, through androids / humanoid robots. Blade Runner is completely built on the protagonist having to interact with rogue artificial intelligence.

> We are in fact very disposable and replaceable.

To your friends and family as well? Or just your employer?

You're describing things that may well be true for a lot of employers, but fall apart outside of that context.