I know we have strong views on what a truly open model is (open weights, open training data, open training code etc.) but I really like how transparent they’ve been about the training of this model.
The realtime dashboard they shared during training (<a href="https://mimo.xiaomi.com/rl/" rel="nofollow">https://mimo.xiaomi.com/rl/) was an incredible learning and teaching tool for me, and they’ve been unusually comprehensive in sharing details about their methodology (check out that tech report - it's got lots of clever behind the scene tricks like Google or Deepseek writeups) and benchmark scores (even the stuff they didn’t do well on).
If you’re releasing an open model going forward, please consider offering the community more of this transparency!
maybe this is why Dario want to slow down AI development and all the big AI labs in the USA is singing the same song.
whey they all singing the same tune. it make me question what is their real motives.
they are afraid of Chinese good enough LLM model killing their margin. we already have story about US companies switch some task to use cheaper Chinese model hosted on Neoclouds.
Please explain how putting an upper bound on how good the strongest models can be prevents cheaper less strong models from catching up, rather than enabling it. I do not understand this argument at all.
The general idea is that Anthropic/OpenAI is pushing this narrative as an attempt at "Regulatory Capture"[1] which would allow them to make it prohibitively expensive for anyone but them to enter the market thus stifling competition.
it wouldn't slow down China as much as make it impossible for American companies to use non-American options, they care about their margins and don't want to be commoditized
...because everyone saw how well that worked for the Jones act, what with all the naval yards the US has lost over time, and how nearly no US-built ships operate where not legally mandated /s
You target the US companies: if they can't use these Chinese models, then they're less of a danger for a now captive audience in the US (and the West generally).
This is already kind of the case: the big enterprises don't really want to touch the latest Chinese models. It's a real pain, personally, I want to use them at work!
Yes, but thats not something a company engaged in regulatory capture for themselves care about: especially if they're worried they'll be outpaced and overtaken by the Chinese labs. Which they will be, IMO.
they care because they know it unlikely open weights will be banned, and thus available to American companies, with regulatory capture (onerous requirements) being a "good enough" "ban" that their big models don't face real competition, regardless of the open weight origin. American companies make open weights too, they are equally threatening to Big Ai financials.
Show them you can burn tokens in seven sessions day and night with comparable results to Opus with less energy and less than 10 dollars a day, per dev.
We have. Unfortunately there are political realities that get in the way, and Bedrock for example doesn't have GLM 5.3 (Flash or otherwise) or anything new/useful
I do imagine it'll change, but it hasn't yet.
If it's hosted, all they know is "data goes to China".
Until profitable, reputable third parties host open models in the US with ZDR or they become plug-and-play for self-hosting at a modest cost, paying the US models is as much about data protection and liability as performance.
I heard someone analogize token vendors to car manufacturers, where American companies only want to produce expensive options, the people want cheaper/better alternatives, and we ban BYD because those with enough money are more "persuasive"
The analogy is a good one, but your explanation is missing one aspect: the country (USA) does have a reasonable interest in having the capacity to build their own models. The “we need to slow down because it’s getting too dangerous” part is probably more related to “we need to slow our public facing development down so the US government can get the best and the American corporations can trickle out what we decide is safe”
It’s similar with cars. It’s not that American cars are better than Chinese cars on any tangible measurement. But America already shipped most of its manufacturing overseas. Everyone who built those factories is retired. The US should probably hold on to some capacity to make cars, seeing as their entire infrastructure depends on them.
American Ai/Car manufacturers could build cheaper/open models, some do, the big ones do not. It's not an either or, but a spectrum where they have chosen to build only in a subrange
That doesn’t explain the decline of German automotive industry which is now taken over by Chinese cars thanks to massive subsidies by the Chinese government
See eg <a href="https://en.wikipedia.org/wiki/Friedrich_Merz#Private_sector_career_(2009%E2%80%932018)" rel="nofollow">https://en.wikipedia.org/wiki/Friedrich_Merz#Private_sector_... for the current chancellor. Many past chancellors were also lawyers, and many members of the Bundestag were and are lawyers.
I don't know whether having lawyers in power leads to industrial decline. My point is only that you can't use Germany as a counterexample.
It's a matter of scale: <a href="https://www.wsj.com/world/china/the-u-s-has-been-spending-billions-to-revive-manufacturing-but-china-is-in-another-league-75ed6309" rel="nofollow">https://www.wsj.com/world/china/the-u-s-has-been-spending-bi...
If you trust Google's AI summary, China spends 4-5% of GDP on industrial subsidies, vs US at 0.4%. 10-12x as much.
Another point of comparison we might make, Trump's desired increase to the US Defense budget is how close to what China is spending on industrial subsidies? It looks relatively close to numbers in these research papers.
It’s not a reasonable comparison. In China, every corporation is de facto state run. The Party is in the boardroom and the executives are members of the Party. The CCP will build entire mega cities or pump money into this industry or that according to their plan.
Local LLM's are hit worse. Its about 6k for 5090 or 15k for an RTX 6000 and the Mac Ultra 256 is upwards of 12k.
Sure if you already have hardware, you can frankenbuild a system - but even the "affordable" dev stations of the DGX Sparks went from 3500 to 5k and upwards of 8k depending on vendor.
All the meanwhile, OpenAI pushed Luna 6 which is crazy cheap suggesting they have flash models to compete with Chinese models.
I just hope we see more open weights. Nvidia has NEMO but their license doesn't allow NEMO to be re-used on say, Apple or ROCm - Nvidia hardware only. Big ol MEH
rao-v · · focus · HN ↗
The realtime dashboard they shared during training (<a href="https://mimo.xiaomi.com/rl/" rel="nofollow">https://mimo.xiaomi.com/rl/) was an incredible learning and teaching tool for me, and they’ve been unusually comprehensive in sharing details about their methodology (check out that tech report - it's got lots of clever behind the scene tricks like Google or Deepseek writeups) and benchmark scores (even the stuff they didn’t do well on).
If you’re releasing an open model going forward, please consider offering the community more of this transparency!
MangoCoffee · · focus · HN ↗
whey they all singing the same tune. it make me question what is their real motives.
they are afraid of Chinese good enough LLM model killing their margin. we already have story about US companies switch some task to use cheaper Chinese model hosted on Neoclouds.
jwolfe · · focus · HN ↗
rbjorklin · · focus · HN ↗
* 1: <a href="https://en.wikipedia.org/wiki/Regulatory_capture" rel="nofollow">https://en.wikipedia.org/wiki/Regulatory_capture
chanakya · · focus · HN ↗
verdverm · · focus · HN ↗
the_sleaze_ · · focus · HN ↗
[dead]
pixl97 · · focus · HN ↗
If the US slows down this may lead to people that would have went to US labs to go to other countries.
RussianCow · · focus · HN ↗
xethos · · focus · HN ↗
eru · · focus · HN ↗
andriy_koval · · focus · HN ↗
girvo · · focus · HN ↗
This is already kind of the case: the big enterprises don't really want to touch the latest Chinese models. It's a real pain, personally, I want to use them at work!
verdverm · · focus · HN ↗
2. Enterprise trends are towards open weights, several routers and vendors now have more than half the volume going towards open weights
girvo · · focus · HN ↗
verdverm · · focus · HN ↗
wolpoli · · focus · HN ↗
verdverm · · focus · HN ↗
His first tweet ever, from this last July
<a href="https://images.nvidia.com/pdf/Open-Weights-and-American-AI-Leadership.pdf" rel="nofollow">https://images.nvidia.com/pdf/Open-Weights-and-American-AI-L...
pimeys · · focus · HN ↗
girvo · · focus · HN ↗
I do imagine it'll change, but it hasn't yet.
unethical_ban · · focus · HN ↗
Until profitable, reputable third parties host open models in the US with ZDR or they become plug-and-play for self-hosting at a modest cost, paying the US models is as much about data protection and liability as performance.
CookieCrisp · · focus · HN ↗
stickfigure · · focus · HN ↗
verdverm · · focus · HN ↗
juiceland · · focus · HN ↗
It’s similar with cars. It’s not that American cars are better than Chinese cars on any tangible measurement. But America already shipped most of its manufacturing overseas. Everyone who built those factories is retired. The US should probably hold on to some capacity to make cars, seeing as their entire infrastructure depends on them.
verdverm · · focus · HN ↗
juiceland · · focus · HN ↗
verdverm · · focus · HN ↗
nxm · · focus · HN ↗
eru · · focus · HN ↗
See eg <a href="https://en.wikipedia.org/wiki/Friedrich_Merz#Private_sector_career_(2009%E2%80%932018)" rel="nofollow">https://en.wikipedia.org/wiki/Friedrich_Merz#Private_sector_... for the current chancellor. Many past chancellors were also lawyers, and many members of the Bundestag were and are lawyers.
I don't know whether having lawyers in power leads to industrial decline. My point is only that you can't use Germany as a counterexample.
verdverm · · focus · HN ↗
In other words, when do economic and industrial policies transition to subsidies? Is it a matter of perspective?
creato · · focus · HN ↗
If you trust Google's AI summary, China spends 4-5% of GDP on industrial subsidies, vs US at 0.4%. 10-12x as much.
verdverm · · focus · HN ↗
verdverm · · focus · HN ↗
<a href="https://www.csis.org/analysis/red-ink-estimating-chinese-industrial-policy-spending-comparative-perspective" rel="nofollow">https://www.csis.org/analysis/red-ink-estimating-chinese-ind...
Some historical analyses of US policies (know less, but both put it over 1% currently, nuances)
<a href="https://www.columbia.edu/~ev2124/research/ErtenStiglitzVerhoogen&Wilse-SamsonJune2026.pdf" rel="nofollow">https://www.columbia.edu/~ev2124/research/ErtenStiglitzVerho...
<a href="https://www.nber.org/system/files/working_papers/w34744/w34744.pdf" rel="nofollow">https://www.nber.org/system/files/working_papers/w34744/w347...
<a href="https://www.columbia.edu/~ev2124/research/ErtenStiglitzVerhoogen&Wilse-SamsonJune2026.pdf" rel="nofollow">https://www.columbia.edu/~ev2124/research/ErtenStiglitzVerho...
verdverm · · focus · HN ↗
juiceland · · focus · HN ↗
supernovae · · focus · HN ↗
Local LLM's are hit worse. Its about 6k for 5090 or 15k for an RTX 6000 and the Mac Ultra 256 is upwards of 12k.
Sure if you already have hardware, you can frankenbuild a system - but even the "affordable" dev stations of the DGX Sparks went from 3500 to 5k and upwards of 8k depending on vendor.
All the meanwhile, OpenAI pushed Luna 6 which is crazy cheap suggesting they have flash models to compete with Chinese models.
I just hope we see more open weights. Nvidia has NEMO but their license doesn't allow NEMO to be re-used on say, Apple or ROCm - Nvidia hardware only. Big ol MEH