Western open-weight models still can’t beat China: Chinese open-weight AI models offer high capability at significantly lower costs than U.S. frontier systems.
14 hours ago by eicker to c/technology
It’s worse than that even… frontier models now have guardrails restrictive enough that I can’t use them for legitimate work. And the ones that DO work have token prices at exorbitant rates, and chew through them like there’s no tomorrow.
Meanwhile, I can download Qwen3 and run it on my own hardware, and it’s about as good as the last generation US model without the annoying guardrails.
Hmm… which model am I going to choose to use?
Can someone explain to me why I, as an American, should be angry that China is making something better than us that it is also releasing for free to the world? I mean, we still get to use it if we want, and so does everyone else.
I don't see why putting more money into the pockets of a few AI billionaires is considered a national priority.
You absolutely shouldn't be angry!
The existence of powerful open-weight models is the single thing that's going to prevent us from descending into the full corporate dystopia everyone's afraid of.
It's going to hit the stock market hard once everyone figures out what's going on though.
I mean, if you have no problems using a tool that says Chinese allies like Russia are the good guys and war against Europe is fully justified, then sure, simply stop caring, just like about Palestine.
Yes, best to stick with non-biased AI like Grok.
implied /s
I wonder if the Chinese are better at it that Musk is, because Grok really will revert back to its training data if pushed just slightly. I'm guessing that, yes, the Chinese are much better at it.
But, really, if you're talking to an AI to form your worldview, you've sort of already lost... that's not really a core use case.
I think the real reason might be that we want AI to make war, so ours needs to be better; and we can't use theirs for our military because maybe there's backdoors we don't know about... which is the same reason nobody in the world wants to buy US military hardware.
Neat! This stuff should not run in data centers anyway. And it should not be gated and controlled by billionares.
I know everyone hates on Meta but they just released an open weight 800B parameter model: Muse Spark Open.
Their smaller 30B parameter open model Muse Glimmer has also been out for a few weeks and is highly competitive with Qwen 3.8 27B.
So I hope we will start seeing more open models from US labs especially in the sub-20GB category which can be deployed on daily use machines.
Trump (and Biden's) legacy.
Open-weight is not the point in the AI race so this comparison doesn‘t tell us anything about who‘s leading in whatever metrics they‘re testing right now.
I wish models had to disclose their training sets so IP holders can sue for damages. Those would get into the trillions no doubt. Hell I wish LLMs never got this much out of hand. But I don‘t understand what they‘re getting at with this or why make it a „China VS the West“ thing.
Open weight AI absolutely matters if the question is who gets to shape the AI ecosystem. A slightly better closed model wins a benchmark: a good open model can be adapted, audited and run without paying a Silicon Valley landlord. And yes, training data transparency should apply to both Chinese and Western labs. That’s not a contradiction: it’s the point. Still: FOSAI » OWAI » ClosedAI
Who owns the compute? The models are irrelevant without the hardware. Leaving you stuck in corpo land.
👌
I’m running a 4090 and barely run 32b and it isn’t half as good as what’s out there on the free tier.
Non starter for 99.9% of the population.
Because then Western AI companies can go to the governments, point to China and say: "see how advanced they are? See how they can apply that to the economy and military? You'd better give us money and not try to regulate us."
That is the whole story.
You’d better give us money and not try to regulate us.
Actually, right now, the western AI companies desperately want regulation.
An unregulated environment is one where they're forced to compete on price against the Chinese companies and startups running open-weight models. All signs point to them losing that battle.
OpenAI/Anthropic's wish list right now is a bunch of regulation to prevent the use of foreign models, and to require expensive to implement "safety" requirements that make it hard for startups to compete.
Yes, that's a different kind of regulation than the one I was thinking about. They do want market/economic regulation (i.e. ban Chinese model use) but not law/justice regulation (our AI made an oopsie, sorry, not our fault).
@lemmy.world
go to feed...
@lemmy.world
go to feed...
Silicon Valley spent billions building AI tollbooths while Chinese labs made models people can actually download and run. Then Washington restricted chips, and China got better at doing more with less. It’s almost poetic: America tried to corner the future of AI and accidentally made efficiency China’s competitive advantage.
save