>I hope announcements like this show that Rust is not a fledgling little language that moves fast and breaks things anymore
I see this on here a lot on this site, but Rust hasn't been that in over a decade. Rust's devotion to post 1.0 stability is massive and has involved some interesting design choices. I started writing run in 2015(?) and only hit one breaking change in the language. It was a niche bug in a macro that was fixed later in a later release.
Just watching hackernews you see lots of news about it, but these are additive and not breaking things. I was still writing lots of mio-style async code after async await was out. You don't have to adapt new style or libraries. I used to have a joke that you could tell a codebase's age based on the error handling libraries used, but even with that it was additive. Often multiple would exist in different parts of the same code base. "Oh wow, I've gone deep on this refactor.... I'm starting to see error_chain"
Hah, I remember error_chain. One of my projects during an internship was upgrading a bunch of the old error handling libraries to the new things. I'm glad that corner of the ecosystem has stabilized now.
At that same job we hit a pretty nasty breaking change where mem::uninitialized() was deprecated and this turned out to cause a lot of critical async libraries to explode at runtime. But these sorts of things don't really happen anymore. The editions system is an excellent design and a big contributor to making the language and stdlib reliable.
You may know more about what it powers. It has enabled much faster, higher quality, more stable security software on Linux. Many modern security tools are built on top of it. While end users might not know it by name, it has made Linux easier to support in the enterprise.
Being the _the_ place for open source is what let them dominate the enterprise. Now that they dominate the enterprise, it'd make financial sense to put hard limits on the free tier. It's cynical but it would be foolish to lose that cash machine. It'd open them up for a competitor to slowly build up good will in open source then take them on in enterprise, but right now they are risking losing it all.
Having worked at places with GitHub Enterprise, that isn’t much of an improvement. You could host something else, but almost everything in this space isn’t great at scale.
Im definitely not referring to GHE. Im talking about real self-hosting infra. It wasnt that long ago that this was possible, and cheaper and more reliable. It's a pay me now or pay me later argument. Everyone now is complaining about the pay me later part of these 'deals'
I don't know how true that is nowadays, I think if you sign up to Enterprise Cloud you're in the same spot as everyone else
(maybe F500 people have their colocated box, but I have the impression that those are the "first class" seats (very expensive but _very few_) whereas enterprise cloud is the "business class" (loooooots of seats making a lot of money)
I'm generally extremely skeptical about a lot of the model hype that show up in comments. Except when there is an extreme mismatch the performance, quirks and quality of these things are difficult to nail down. You wouldn't know that from the comment section of every single release.
I think some of these are excited, eager users always ready to hype up the new thing. The same crowd that previously would constantly push for a rewrite from angular->react->svelt->god knows what. Instead now it is on a 6 week cycle and about models/harnesses.
I think most of it is bot driven spam by the various labs. It's hard not to notice 3 month old accounts with very strong opinions about various frontier labs and little else.
Ultimately, I think some of it is legitimate shifts in who's in lead and what is the best. You gotta dig through a lot of crap to get to that, and I don't really know how to to.
Ultimately I'm saying is that I always applied a fair amount of skepticism about what I see in comment sections but these day it is extreme amounts.
> I think most of it is bot driven spam by the various labs. It's hard not to notice 3 month old accounts with very strong opinions about various frontier labs and little else.
I've assumed the same as well.
I also assume that many of the companies developing these models engage in benchmaxxing.
At my company we've developed our own internal benchmarks for evaluating LLM models as they become available. The benchmarks are tailored to our particular use cases but the utility and knowledge our benchmarks assess is still fairly universally applicable. I see wide differences between what our internal benchmarks report and what the major benchmarks do.
There was a whole lot of fanfare about how amazing GLM 5.2 was when it was released, but it was pure rubbish on our internal benchmark -- far behind OpenAI, Anthropic, Gemini, DeepSeek, etc. I don't know how to reconcile the fact that GLM 5.2 performed very well on some of the major public benchmarks, but consistently performs so poorly on ours. OpenAI models tend to dominate our internal benchmarks.
It actually is showing in public benchmark if you know how to look for it. For example, in Terminal-Bench 2.1, GLM 5.2 received 78%, while GPT 5.6 Sol received 88%.
Then Terminal-Bench 3.0 came out (where the questions are new), and GPT 5.6 Sol received 34.6%, while GLM 5.2 dropped to a whopping 4.6%.
> OpenAI models tend to dominate our internal benchmarks.
That's odd, since Fable seems to be the leader for the industry. Not cost-effective, but if Anthropic models get dominated by OpenAI in your internal benchmarks, this calls their validity into question. Separately, see the jagged frontier effect. [1]
I've been using GLM 5.2 at my day job (mostly Rust backend work ATM). Nothing that blows away the models from OpenAI and Anthropic, but solidly good enough to get it done. A lot of people have experienced this and the fact that an open weights model can do so is where most of the excitement comes from. Optimizing for benchmarks can only get you so far, and people are quick to criticize models that fall into it (like DeepSeek Pro V4 recently).
I think most of it is bot driven spam by the various labs. It's hard not to notice 3 month old accounts with very strong opinions about various frontier labs and little else.
Really no different than when there was suddenly online personas everywhere hyping up TSLA out of the blue. You can see the same thing going on with the BoringCompany subreddit. Crazy that the botnet master isn't able to convince us that Grok is also the best model. I don't think buying twitter was an accident it was probably just literally covering up the evidence.
IMO this is getting hyped because the 27b version runs on a decent gaming GPU. This is NOT a thread for their largest model, this model will run on a mid-high end gaming PC, which you probably have in your household. Mine is 6 years old and it runs quite well.
I strongly suspect that many of these accounts you think might be bots from the labs are just people who only have one interest.
One thing I have noted a lot more of is that there is comparative fanboying going on. Like "this model has done badly, my favoured competition has a model out soon that will beat this in every way" — comparing a released product to unverifiable hopey claims about an unreleased product.
It's tempting to assume that is bot stuff, but if you've been around any other "hot" technical hobby online (cameras, phones, 3d printers, whatever) you will know it's not. It's just fans aligning into teams, some of them laconic and amusing, some of them overkeen and toxic.
It often is. Risky Business Features did a fantastic podcast on how different popular methods of guardrails work and some popular methods on defeating them. Absolutely worth a listen because there are some surprising insights in there on how these work, even for day to day use, not just bypasses:
Even more than supporting any agent that support almost anything period. They seem to take the default approach of using open community standards first when adding new features. By investing very heavily in LSP, ACP, DAP, tree-sitter etc it makes it easy for Zed to integrate with tons of different workflows, language, ecosystems etc quickly and easily. Its just good engineering and I love to see it.
> An unprivileged userspace application, even if coded poorly, shouldn't crash the machine.
How would you prevent that? It can write until root disk is full, can take all your memory, saturate your CPU and a whole bunch of other stuff we (users) like to be able to do for the things that need them.
Unless you're suggesting something like Qubes or similar UX?
Right, my point being that it's should be up to the user to decide, because most of the times I run programs, I actually want them to be able to gobble up GB of disk, RAM and fully make use of my CPU. That's why I bought the hardware I have :)
Isn't that the default? The user has access to all available memory unless they (or an admin) set a restriction. Unavailable memory is only unavailable because writing to it crashes the kernel, thus preventing you from using your hardware!
Yes, it is, and for good reasons. That's why I disagreed with the original "An unprivileged userspace application, even if coded poorly, shouldn't crash the machine" part.
I don't follow. The default today is that an unprivileged user space application should not crash the machine. The kernel reserves what it needs, then hands the rest to user space. Thus, all available memory should not crash the machine.
I can't wrap my head around the idea that distillation is IP theft but mass training on books, music and art without consent is fair use. The two stances are incompatible. If it is transformative use of a book, it is transformative use of AI output.
Yeah people in the US who want protectionism for US models on grounds of IP rights are appalling hypocrites. What’s good for the goose is good for the gander!
I really hope that when a sane administration returns to power in the US we'll actually get a reckoning over how irrational it was to not restrict training data.
The idea is that selling, or giving away art grants the consumer to sell art of his own, not a clear copy of.
So the court rationally found Anthropic guilty for acquiring copies without paying for them. But didn't charge the training, deemed fair use.
If it isn't fair use, then we may need to sue all teachers for spitting out knowledge they ultimately acquired from someone's work.
Copyright, some would say is irrational. Humans learn, that is copying, distilling in fact.
Copyright though is pragmatic: it draws a line, art will be reproduced, let's just enforce that they can't be shameless (near) identical copies.
Mix it up, derive the original enough so that you aren't competing with the original author.
The only argument to ban training on copyright data is that it unfairly compete with original authors. Which stance do you take? Neither would be irrational.
I agree that the fair use argument on training data is a genie that’s not getting put back in a bottle. But:
> If it isn't fair use, then we may need to sue all teachers for spitting out knowledge they ultimately acquired from someone's work.
Is such a lazy argument and always was. A human doing something is necessarily different than a machine doing it. We can be ok with a human doing a thing and simultaneously not ok with a machine doing it.
I side with the argument for machines learning being treated differently than human learning.
And in doubt, to pause and forbid commercialisation given the clear impact it's already having on people's financials, humans have invested in a given set of rules that ML are disrupting.
My argument is simply that the ruling for fair use isn't irrational. It's a rationality that you, me and many others would disagree with. Lazy? Yes. Also call it lazy, and biased towards VC's interest who can't squeezed much profit in status quo.
> I suppose you think that model providers are not allowed to impose restrictions on the use of their model
They should have exactly the same ability to "restrict" it is as the writer of a book on those who read the book that was purchased.
And seeing as those model providers were not restricted from training on those books.... It would follow that other people would have the exact same right to train on the output of the models.
I simply demand that model providers are treated exactly the same as the data that they trained on. Either it was OK for them to train on other people's work, en mass, without permission, over the objection of the creator, and therefore its OK to do the same to them.
or none of its ok, and they should presumably be equally sued into oblivion, and equally shut down completely by the government.
Thats all. Take your pick. Either all of the training on either books and all the models, without permission, is ok or none of it is.
Additionally, the output of a model isn't even copyrightable. So actually there would be even less protections for that. Because of this, it seems that anyone could use it for anything.
EX: 3rd parties aren't bound by the TOS of the models. So someone could simply do a passthrough, and give the uncopyrightable output to someone else to distill, and since the distiller didn't sign the TOS they would be in the clear to train on non copyrightable info.
I am saying that someone else would use that output to create or improve a different model. I think you could have figured out that this was the meaning of my statement instead of doing the irrelevant nitpick that you did.
Its also unrelated to my point, which is that person 1 could give the data from model A to person 2, and person 2 would be fine because they didn't sign any TOS contracts with model A, and the output from model AI is not copyrightable and therefore can be redistributed.
Additionally, it still doesn't address the point about how the original model trained on a bunch of other people's stuff without permission, so I don't see why the same shouldn't be done to their outputted content.
Do you have any substantive disagreements or are you just going to make a minute, incorrect nitpick and then not elaborate?
I'm glad someone ran the numbers on this. Every single Simon Willison post of an SVG is followed with someone dismissing it saying "I'm sure they train on it by now." This is despite a good blog post with sound logic on how easy that is to catch. [1] Glad to see someone took the time for a quantitative analysis of dumb little animals riding dumb little bikes.
I don't think this small amount generalization to other animals and vehicles is strong evidence they haven't trained on this, either directly or more generally.
Honest question how could they possibly train on this as there are no good SVG pelicans to train off right? So they’re just training off a bunch of bad ones which should lead to just bad pelicans, but the pelicans are getting better.
Training on generating SVGs directly at all is already fairly niche. Generating full scenes with a cartoony character is even nicher. But there's plenty of non-pelican cartoony SVG content out there (created, not written, by humans with vector design tools), and more importantly, plenty of vision models to give feedback on the output (just raster as a png). You could easily hill climb this niche skill, if you cared.
I agree with the conclusion and am happy to see this blog post, but this killed a bit of credibility for me:
> Using a single LLM judge for scoring. Every score here comes from one model, GPT-5.6 Luna, looking at one image at a time. I didn’t do much alignment and didn’t check how often it agrees with itself on a re-run.
Having used a similar setup (with previous gen LLMs) to evaluate the 3D models that my product[0] generates, it turned out there was no correlation at all. LLM judgments were very much random and I assume judging SVGs is not that far from judging 3D models. I guess I have to re-test this with current gen.
There's that version of the argument version, but there's also the softer version: that there used to be no training material of illustrated pelicans on bicycles, but now you have actual artistically talented individuals drawing it and that could improve the performance even though the AI labs are sucking it up no differently than everything else.
This post proves that hasn't happened yet, either. Although maybe the bad results posted online are being trained on and that explains the UNDER performance.
If you remove the word "AI" from your post it also describes the last 20 or so years of US/Chinese economic competition. They are real, serious competitor but also there is a lot of value in that last 10%. Some buyers need the best and you can charge a premium for it.
The speed in which a copier can catch up to a leader is faster than every before, but everything is faster these days. I think the threat is if can these labs catch up and pass American labs while keeping costs low, but I don't see that happening.
>Some buyers need the best and you can charge a premium for it.
Yes, and that is how you get a top-heavy k-shaped economy I believe. It is harder and harder to sell premium A+ products, because most people can only afford what works. If my old ice cream vendor charges me 25% more but puts chocolate sprinkles on it i'm going for the non-sprinkle version.
> the last 20 or so years of US/Chinese economic competition
Only with some industries, mostly electronics and light manufacturing. The rest of the history of software products doesn't look like this at all. China cloned Amazon and Meta fairly well (but only within their local/captured markets) but not Microsoft or Google and absolutely not Apple. Entertainment and culture exports are still effectively zero (like, what, three blockbuster films?) where the US, Japan and Korea define pop across the globe. Heavy industry is mixed, with decent success in consumer vehicles now but little purchase in stuff like heavy equipment or shipbuilding.
Basically it's complicated, and not really a point well suited to that kind of generalism.
> China cloned Amazon and Meta fairly well (but only within their local/captured markets)
Americans order Chinese products from SHEIN and Aliexpress type websites.
So China did penetrate the US ecommerce market, but they are not taking on Amazon directly.
People are buying from the Chinese and reselling on Amazon, which has a good logistics network that Chinese companies can't possibly build in the USA.
> absolutely not Apple
You need to look up Oneplus to see how many people were using their devices till they announced exit from Western markets.
The Chinese Android phones are essentially replacement for a big chunk of Apple ecosystem.
If these cheap mobile devices didn't exist - Apple could possibly be selling many more devices.
> China cloned .... Meta fairly well
TikTok is giving Meta a run for their money in USA.
They had to be restricted via threat of regulation to offload stake in TikTok USA.
> Microsoft
They didn't compete with Microsoft directly as entering Western Enterprise market is hard for the Chinese.
Even Google can't beat Microsoft there till now.
Oneplus is owned/backed by Oppo. Their other phone brands like Oppo, Realme, etc. are really popular worldwide. They removed Oneplus from some regions (e.g. EU) where some of their other brands are more popular.
I'm not sure if you are from the US, but if you are, it really gives you a distorted image of what the worldwide smarphone market is like.
E.g. in many European countries, it is 1/3 iPhone, 1/3 Samsung, 1/3 others (which are mostly Chinese brands).
Not British, but taking UK out of the definition of pop, is a bit too rough a cut. Also France. Look at the queues outside pop-glamorous stores. And was pop-art a pop artefact? Is TikTok redefining pop? Movies is an old measure, and even US has fallen with just franchise dominating. I think as a globe we are much more integrated and co-dependent, occasionally also in bad ways, than we appreciate.
I'm talking about economic activity more than the passports of the individual artists. And no, TikTok is a software product (a social media platform), not popular culture. There's absolutely popular culture being created and driven there but it's 100% NOT coming out of China. Are you high, as it were?
TikTok competes with Meta, not Olivia Rodrigo or BTS.
I am not high, I have kids, and can tell you they barely care about Olivia Rodrigo or BTS. You dispel passports as irrelevant for one leg but base your argument on them for the other. The contradiction is sufficient.
> Entertainment and culture exports are still effectively zero (like, what, three blockbuster films?) where the US, Japan and Korea define pop across the globe.
And this is really sad, because they have thousands of years of cultural heritage and also plenty of great films. Somehow, they've failed at exporting this completely.
> Heavy industry is mixed, with decent success in consumer vehicles now but little purchase in stuff like heavy equipment or shipbuilding
Uhh, China does over half of the world’s shipbuilding.
China is also responsible for 25% [1] of the world’s heavy equipment manufacturing and is considered to have double[2] the CAGR of the global average for heavy equipment manufacturing.
the difference is that the US economy is currently over-investing in AI to the point where it's a significant portion of the economy, it's the biggest competitive opportunity in quite a while
I see this on here a lot on this site, but Rust hasn't been that in over a decade. Rust's devotion to post 1.0 stability is massive and has involved some interesting design choices. I started writing run in 2015(?) and only hit one breaking change in the language. It was a niche bug in a macro that was fixed later in a later release.
Just watching hackernews you see lots of news about it, but these are additive and not breaking things. I was still writing lots of mio-style async code after async await was out. You don't have to adapt new style or libraries. I used to have a joke that you could tell a codebase's age based on the error handling libraries used, but even with that it was additive. Often multiple would exist in different parts of the same code base. "Oh wow, I've gone deep on this refactor.... I'm starting to see error_chain"
reply