HN user
And I know many people that don't. That have a 20 or 100 dollar subscription for very bursty workflows with months where they barely use tokens.
Not every subscriber is a full time SWE. In fact most professional SWEs will be on enterprise plans and thus not get subscriptions at all.
I think it's very plausible that subscriptions are overall losing money. But we simply don't know.
This is hotly debated and completely unclear. Let's say Anthropics Opus models cost the same to serve as GLM 5.2. GLM 5.2 is 4.4$/MTok while Opus is 5.6 times more expensive. Assume that GLM 5.2 is served at essentially zero margin. Then Anthropic has >80% margin on API pricing. So even if an average person with a subscription pays only 20% of the API price of their usage, Anthropic makes money on subscriptions.
And the real numbers could be better for Anthropic. It's feasible Opus models are actually cheaper to serve than GLM 5.2 because Anthropic have optimized the hell out of inference.
Please post an example of a response that shows such masking.
* * *
I read the article you linked:
https://www.chinatalk.media/p/how-to-buy-cheap-claude-tokens...
I know that Anthropic claims that "large-scale coordinated "distillation attacks"" exist. They are not a disinterested party here and they don't provide evidence either.
Also you're just moving goalposts around. The article you linked also cites Anthropics claim but then paints a much much mor nuanced picture.
This topic really is a testament to people's willingness to opine on things they have absolutely no clue about.
A first year undergraduate can completely check this counterexample in ten minutes. The original post even linked Wolfram alpha for the calculations.
And if you genuinely try you can very quickly understand using only high school math and a bit of Wikipedia that this counterexample is vanishingly unlikely to be wrong, even if you don't do the calculations yourself.
I mean, China is the leader in PV panels, battery tech, telecommunication hardware, etc... already...
First off, I don't doubt that China engages in large-scale industrial espionage, or at least used to. Nowadays they have plenty of talented and highly educated engineering talent of their own.
Second, I now actually read the article. It describes plenty of questionable and problematic things but also contradicts your claims explicitly.
The essential point of the article is about making money by selling access to Claude cheaply in China. Not about Chinese labs orchestrating a way to get their hands on Claude output.
Your credit card claim is considerably weaker in the article: "[Beyond this there are] accounts purchased using stolen or fraudulent credit cards [...]. How large this share is relative to the above four “innocent” tactics is difficult to verify, but the two markets likely share some infrastructure and personnel.". Instead, swapping models to cheaper alternatives is listed as a major reason for cheaper prices.
Then the article gets a key point wrong: As many others have pointed out to you, you don't get access to the reasoning traces anymore on the subscription accounts. And the article also clearly states:
"Chinese developer communities assert [selling logs] is happening in at least some cases, but whether proxy operators are systematically harvesting and selling these logs, and to whom, remains unverified. However, downstream distillation data does exist on the open web. Several datasets of Claude Opus 4.6 reasoning outputs circulate on HuggingFace with no clear source for the outputs. Theoretically, one can clean and sell similar distilled datasets to other model developers in China."
The article also discusses selling logs for other (far worse!) purposes than for training, like blackmail.
So overall this article reads very, very different to your claims. Nothing in the article suggests or supports the idea of large-scale coordinated "distillation attacks". Instead it paints the picture of a naturally emerging grey-market response to access control blocks, consisting of many exchangeable individual actors: "Almost no one operates the full chain. Most participants own one or two links and monetise those well, resulting in a resilient, modular system."
Importantly: Nothing in any of this looks ethically worse to me than Meta using pirated books for training. And nothing suggests that OpenAI or Anthropic were more ethical than Meta when sourcing their material.
TOS violations are not espionage. Everybody who links up Claude to OpenCode is violating the TOS.
So from the largest industrial espionage in history we have left "They paid for the accounts but violated the TOS". And then you randomly add the claim they stole the money to pay for the accounts.
You have provided no evidence other than "Claude tokens are sold for cheap in China". As others have pointed out, that might also simply be counterfeit tokens generated by open weight models.
The western labs have established the precedent that all data they can buy beg borrow or steal is fair game. Turning around and crying foul when the Chinese labs follow their lead is hypocrisy.
You haven't explained how this is illegal or any more immoral than scraping the web for training data.
As you said yourself: They are buying the product. Then they are using it for their own purposes. That's more than Anthropic/OpenAI did for the open internet. That's more than Meta did when they obtained torrents of books in the early days, and then claimed that even though the data was obtained illegally they can still train on it just fine.
They paid for it! It's absurd to call this espionage!
Can I interact with and discover any federated instance without having to know it exists?
My experience with Bluesky Vs Mastodon really showed that the friction of federation in the latter can really kill the experience for me. I think we need something like Signal is to WhatsApp but for GitHub and my impression is that the ATProto world is the only one with the potential to deliver this.
In my reading, "training for the benchmark" is very, very different from "this benchmark is in the training data".
Fire predates modern humans. I will concede agriculture as up there with the two I mentioned, but I think you underestimate how radical the shift of industrialisation was/is, and I think this is because history teaches us about highly atypical lives from the past. City dwellers, kings, etc...
In fact you could reasonably argue that life for most people didn't change that much from establishing agricultural settlements all the way up to industrialization. Most people spent most of their time procuring food and clothing/shelter using roughly similar methods.
Not the person you are asking, but... In my view, the two most important technology clusters in human history are writing and industrialization. The former made culture (cognitive work) cumulative. The latter made physical work approximately free, with the result that we now have 5% instead of 95% working on producing food. Now we have a technology that could potentially be a culmination of these two: The industrialization of cognitive work/culture.
It isn't absurd to think this is one of the most consequential technologies ever. It probably doesn't make sense to rank technologies (which build on top of each other after all) by importance, but it's fairly clear this is a historic breakthrough.
Here is what Sonnet says on genocide:
"Surveys of genocide scholars (e.g., one by the journal Journal of Genocide Research contributors) show meaningful disagreement, though a visible and growing share of specialists in genocide studies specifically have concluded the term applies or is defensible — more so than international law scholars generally, who tend to be more cautious about the intent requirement."
You can be of the opinion that this shows bias, but it's a far cry from the Chinese censorship.
Clichés like "freedom is relative" are not serious arguments. You can't genuinely argue that the US, for all its problems, is less free than China. Words have meaning.
E.g. looking at freedom house index you can disagree on the precise score or how different aspects are weighed, but you can't argue with the underlying facts in the country reports
That's the idea behind the future file format:
https://github.com/future-file-format/F3
Bundle a wasm decoder as a fallback when native decoder isn't available.
That's a strange statement... It's been true for a while now that OpenAI has had much more generous limits than Anthropic on their subscription plans. And with the Fable ban/guardrails disaster, there has been a lot of frustration from people in these comment sections. And Anthropic fucked up Claude Code pretty badly for a couple of weeks during the 4.6/4.7/4.8 transition, which again was widely publicized. And they got a lot of flack over not allowing other harnesses anymore. And ChatGPT got some pretty viral wins on model intelligence when they cracked the high profile Erdos problem.
If anything the online optics have been bad for Anthropic for the last half year. OpenAI doesn't have optics issues, from my point of view they simply have the issue that they are the least trustworthy player at the frontier. The way they pivoted from their original mission is truly breathtaking, especially coming in gloatingly to take the government contract when Anthropic got kicked out for insisting the government does not use their systems for mass surveillance or autonomous weapons systems. You understand what that means, right? OpenAI models are now actively used/developed for mass surveilance and/or autonomous weapons systems.
I know there are plenty here who seem to value their own ability to use these models cheaply above all other considerations. Then OpenAI is a great choice, and much less restrictive than Anthropic. But their problem is not on the optics. It's on the substance.
No. Robust control theory was one case. Dynamical systems another.
You are mischaracterizing what the post is reporting entirely. Porting an open source tool used in bioscience to rust is a software engineering task. But it is somewhat understandable that it gets stuck in the overly broad safety margin.
But I do research on stuff that is entirely unrelated to bio or cybersecurity, and the model is simply not taking any of my research-level prompts. This is fairly abstract mathematical stuff. All of this, including all the examples in the posted article, are far from "trying to get the model to do what it was made not to do".
I disagree. When I got Fable to engage with research questions before they tightened the guardrails it was a genuine step up from Opus 4.8. I see no real reason that what everybody reported isn't exactly what happened.
With these guardrails it is completely useless. The only hope is that they eventually convince the US Gov to let them use a saner classifier.
My experience, too. I work on nothing in any way related to cybersecurity or biology. I asked it a few purely mathematical questions, it refused immediately.
Before the export embargo I did get it to look at some hairy problems and the output was genuinely useful...
What is your point? Sometimes it works due to other factors so we don't need regulation?
Beyond the very real closures of unprofitable rail lines that is always happening or threatening (including in state owned rail companies) your examples don't really measure up either:
Local airports are typically subsidized. So the infrastructure aspect is not market driven.
Eventually getting starlink means it was fine to not have fast internet for a couple of decades? We still don't have good mobile internet on many train lines in Germany. In Poland it's fine. The difference? Poland mandated that companies bidding for 4G meet certain coverage goals. Market forces have absolutely failed to actually get reasonable mobile internet to everyone.
There are many more places that were never reconnected.
Officially it's not available anymore. But there are team plans that are not enterprise, so if you are small enough and fine with the data protection included in those maybe that is what you're working with? And I do know NGOs were offered seat based plans after they were officially not available anymore.
Also: This change came in in March so if you got your contract before then this will only bite once you renew.
https://support.claude.com/en/articles/9797531-what-is-the-e...
Individuals, sure. For enterprise you can't get monthly plans. You have to pay per token.
It's a bit like saying "nobody pays for Microsoft Office". I certainly don't know anyone personally who has. Students get a free Education License and then your employer provides one for you...
Infrastructure should not be (purely) profit driven. To improve profits for train operators, the simple option is to cut lines serving small and rural communities. The economics are much worse than serving large cities. Same for cell coverage and broadband internet. Most profitable is to just not cover a few percent of the population.
There is a point where technology becomes foundational for participating in society. And then it needs to be regulated to be available to everyone.
The question is: what proportion of tasks can not be handled by GLM5.2?
How many software developers were working on code like the one you describe?
So requires more transmission infrastructure. The difference is that we already have that built out over decades, and now we need a different network in a much shorter timescale.
No one should pretend that the energy transition is free. The final system we will arrive at can be ver
https://www.climate.gov/sites/default/files/2022-10/global_t...
CO2 and temperature track quite well. However, climate sensitivity actually says the relationship is not linear but logarithmic. Doubling CO2 increases temperature by 2-3 degrees.
The thing is, in the last decade or two we have firmly moved into the regime where we are out of the natural variability. If you get a 30 degree summer every five years instead of every ten that might be a very clear signal for warming, but is not as notable. If every summer is 30 degree and sometimes you get 40 you really feel the new climate normal.
"Carl Benedikt Frey at Oxford has documented that the Industrial Revolution took seventy years before wages and employment recovered for the workers it displaced. In the interim, wages stagnated, the labor share of income collapsed, profits surged, inequality skyrocketed, and the political consequences included the Chartist movement and widespread social upheaval."
From half way through this (meandering) blog post:
https://www.owenmcgrann.com/p/the-dead-economy-theory
As for the populism link, that is well established empirically:
https://academic.oup.com/oxrep/article-abstract/34/3/418/504...
https://www.iza.org/publications/dp/12485/we-were-the-robots...
Etc...
Edit: Just found this when looking up sources, I haven't had time to look at it but just dumping it here:
https://www.cambridge.org/core/services/aop-cambridge-core/c...