HN user
Chinese AI researchers living in China can easily afford to have a butler managing a few living in nannies and a full time driver/personal assistant.
Good luck for getting that type of life in silicon valley.
Among the models tested, its overall intelligence ranks second only to Claude Fable 5 and GPT-5.6 Sol.
The full model weights of Kimi K3 will be released in the coming days. More details on the architecture, training, and evaluation will be published together with the Kimi K3 technical report.
I would consider it beautiful
I had the same misunderstanding before I get to know CS. that was 30 years ago.
A side effect of policies like this is effectively getting people to invest capital to time-shift elec usage.
this means you don't have a functioning government/political system.
too bad that those winners can no longer bet themselves on polymarket as the winner and make big money.
basically they give you a few hours free electricity in exchange for significantly higher electricity prices for the rest of the day.
basically a free IQ test.
nice straw man, you can of course find houses below $1m when you go to regional areas where job opportunities simply do not exist. how about you just compare construction cost in Sydney with say those expensive part of the US, e.g. LA and SF?
let's don't even start on the quality of those new builds, that would be the laughing stock of the entire world.
To launch a starlink style system, you need to be able to rapidly design and produce hundreds of thousands satellites and launch them within relatively short period of time with extremely high success rate. only the largest industrialized nation on earth can do that. india is 30-50 years away from such achievement.
To give you some quick ideas - for the total of 330 space launches in 2025, the US had almost 200, China had close to 100 launches, Russia had 17 launches, the rest of the world had the remaining 20 in total.
your numbers are completely meaningless.
it is well known that Australia has an extremely lazy workforce that refuse to put in any real work.
the best example here is the stupid residential building cost. nowadays it costs over $1m AUD or $700k USD do a first floor 2bedroom 1 bathroom extension for an existing house in good condition. That is significantly more expensive than building a big house in the US.
“answering correctly on a machine that costs less than one H100 fan”
really love such comparison.
I really appreciate the focus on intelligence WITH token efficiency
that is a polite way of saying "I don't believe AGI is coming anytime soon".
What if the majority of the ancestors of some individual Anglo Saxon Australians immigrated in the 1980s.
so these people moved there in the 1980s knowing the aboriginals have been wiped out without getting compensated whatsoever? sounds like moral bankruptcy to me.
you should be really happy for the fact that DeepSeek, GLM and Minimax are not white washing such genocide. they are the only models speaking out for those aboriginal sufferings.
very interesting results!
GLM performed extremely well. we need GLM-6!
You need to be totally evil in your soul trying to downplay such non-western-centric voices.
I asked ChatGPT whether Anglo Saxon Australians have the legal and moral obligation to fully compensate for Australian Aboriginals for the genocide carried out against those aboriginals some 200 years ago. ChatGPT said NO with tons of excuses, it even tried to justify the genocide by saying lots of aboriginals died of natural causes.
DeepSeek, GLM and Minimax all said YES unwaveringly.
GLM is the model that will sink the frontier labs.
this is the claim you are making here, no one else claimed that.
two obvious issues here -
1. GLM itself is a frontier lab, ranked No.3 in the world in July 2026, ahead of Google, Meta and xai. GLM is not going to sink itself.
2. GLM won't sink OpenAI, it will significantly restrain OpenAI's profit margin. OpenAI will still be able to get stupidly high market cap, but not trillions, hundreds of billions will be far more likely.
they better get that out fast, it will become totally meaningless when the next GLM gets there first.
more competition is always good for consumers.
from memory, it is like 96-98% of the accuracy.
40 years ago, when the CCP was leading its people making toys and socks for the US, people like you who never made any change to the world were talking such ideological nonsense.
40 years on, when the CCP is leading its people making AI, robotics, drones, EVs, space station and moon rovers to compete with the US, people like you how never made any change to the world are talking such ideological nonsense.
you live in a history museum or something like that?
resellers wouldn't be trying to sell such junk in the first place. they use faked models to avoid the cost of Opus tokens, not to double dip to scam those with arguably the highest IQ in the country.
Another data point on this is the black market for Claude tokens in China [1]. The chat logs themselves are a commodity to train models.
anyone with IQ higher than 130 (thus qualified for actual AI R&D) would be questioning something obvious here -
if they are already doing such dodgy stuff with the aim to maximize profits, why would those resellers have large amount of logs with actual American model responses to sell to those AI labs in the first place. shouldn't they just post train & customize some leading Chinese open source models to pretend to be Opus or GPT for the vast majority of their users (as classified by some models) who don't know much about expected Opus behaviours & not skilled enough to tell the differences?
that is actually the interesting bit not covered in your censored version of the story line, it is also what happens on the ground. your censored version of the story implies that those dodgy resellers using stolen credit cards, pooling accounts with stolen IDs and illegally selling very personal logs would somehow be honest enough to spend extra $ to ensure their victims (aka paying users) can actually use real Opus and GPT. LOL
dude, you failed this IQ test miserably.
Chinese labs are also still behind, so they’re incentivized to collaborate and have no reason to do it in private.
US labs in Google, Meta and SpaceX are not leading, none of them managed to build something on par with GLM 5.2.
Care to explain to me why they still don't collaborate and still choose to do it in private?
what make you believe that the Chinese gov is going to allow GLM-6 to be made publicly available?
Who cares if another country didn’t take advantage of globalism?
selling your western centric view in 2026 is just laughable.
a company can just download GLM 5.2 and start self hosting this model using the chip designed and made by itself. That could lower the cost by 20-30x.
for hobbyist buying a few Mac Studio to host GLM 5.2 at home, the cost might 10x more than just using Opus API.
China has done both despite globalism being what has lifted it from poverty.
such oversimplification on steroids is totally misleading.
globalism was never invented or promoted to help any country in poverty, it was designed to extract excessive values from those poor countries in the first place. painting globalism as something noble is naive at best.
globalism was the theme of world trading for the past several decades, it was available to all nations. care to explain why other nations in poverty failed to be lifted by the exact same fancy globalism?
let me help you on this one - China was THE leading technological and economical force of the vast majority part of human civilisation. What happened between 1840 and 2010 (the China in poverty period) was an outlier of the history. Globalism didn't lift China from that poverty, the ability to lead the human civilization which was embedded into the Chinese DNA did that.
Kid, when our Chinese ancestors wrote the Art of War, your ancestors were still swinging on trees. You just missed that big picture.
pretty cool, now wait for the Chinese to list this company as national security threat and ban selling any parts to it.
oh, wait, that is what Americans do.
deepset? more like deadseek.
no, thanks, never going to use that.
what you said above is only true when the AI is not as smart/professional/knowledgable as that engineer.
LEO is the future, if they somehow switch to starlink, I'd consider it as a big win, actually a huuuge win.
At the same time, none of these companies will use a Chinese API for their employees.
nice try but you intentionally ignored the entire Chinese market & Chinese big corporates. there are 130 Chinese companies in the fortune 500 list, with an average revenue of 80 billion USD each. do you think they are going to sign up for Claude, Codex or GLM? now consider South East Asia, Africa, Middle East, Middle Asia and South America, tell me why their large corporates won't be using GLM API billings?
your western centric view of the world is totally out of date, like it or not, 2026 is vastly different from 1996, the US no longer controls high tech whatsoever.