Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Qwen3-Next

qwen.ai

Post reply on HN
Print view View original

171–180 of 240 posts

Re: Qwen3-Next

#171

Post by moffkalast »

NitpickLawyer wrote: post #59
moffkalast wrote: post #43

Earlier quoted context omitted.

In retrospect it's actually funny that last year Meta spent so many resources training a dense 405B model that both underperforms compared to models a tenth its size and is impossible to run at a reasonable speed on any hardware in existence.

It's not that clear. Yes, it underperforms in recent benchmarks and usecases (i.e. agentic stuff), but it is still one of the strongest open models in terms of "knowledge". Dense does have that advantage of MoE, even if it's extremely expensive to run inference on. Check out this great exercise - https://open.substack.com/pub/outsidetext/p/how-does-a-blind...

Ok wow that is incredibly interesting, what a test. I would've honestly expected just random noise (like if you gave this same task a human, lol) but you can even see related models draw similar results. Maybe it is an indicator of overall knowledge, or how consistent the world model is. It also could not correlate at all with non-geographical knowledge.

Re: Qwen3-Next

#172

Post by bigyabai »

jwr wrote: post #44

Hmm. 80B. These days I am on the lookout for new models in the 32B range, since that is what fits and runs comfortably on my MacBook Pro (M4, 64GB). I use ollama every day for spam filtering: gemma3:27b works great, but I use gpt-oss:20b on a daily basis because it's so much faster and comparable in performance.

The model is 80b parameters, but only 3b are activated during inference. I'm running the old 2507 Qwen3 30B model on my 8gb Nvidia card and get very usable performance.

Re: Qwen3-Next

#173

Post by bigyabai »

EnPissant wrote: post #41
regularfry wrote: post #39

Earlier quoted context omitted.

This isn't quite right: it'll run with the full model loaded to RAM, swapping in the experts as it needs. It has turned out in the past that experts can be stable across more than one token so you're not swapping as much as you'd think. I don't know if that's been confirmed to still be true on recent MoEs, but I wouldn't be surprised.

What you are describing would be uselessly slow and nobody does that.

I run the 30B Qwen3 on my 8GB Nvidia GPU and get a shockingly high tok/s.

Re: Qwen3-Next

#174

Post by davidpfarrell »

Hyped for the release, but bummed they fell for the ‘next’ naming convention.

What will the actual next advanced release be called:

* next-next

* next (2)

* actual-next-final

Re: Qwen3-Next

#175

Post by amelius »

eMPee584 wrote: post #167
amelius wrote: post #121

Earlier quoted context omitted.

> if that were infinitely scalable, wouldn't we have a subset of super-smart models already at very high cost The compute/intelligence curve is not a straight line. It's probably more a curve that saturates, at like 70% of human intelligence. More compute still means more intelligence. But you'll never reach 100% human intelligence. It saturates way below that.

how would you know it converges on human limits, why wouldn't it be able to go beyond, especially if it gets its own world sim sandbox?

I didn't say that. It converges well below human limits. That's what we see.

Thinking it will go beyond human limits is just wishful thinking at this point. There is no reason to believe it.

Re: Qwen3-Next

#176

Post by jug »

seunosewa wrote: post #73
jychang wrote: post #60

Earlier quoted context omitted.

Strong disagree. Llama 4's release in 2025 is (deservedly) panned, but Llama 3.1 405b does not deserve that slander. https://artificialanalysis.ai/#frontier-language-model-intel... Do not compare 2024 models to the current cutting edge. At the time, Llama 3.1 405b was the very first open source (open weights) model to come close to the closed source cutting edge. It was very very close in performance to GPT-4o and Cl…

He is definitely talking about Llama4.

Llama 4 is neither from last year nor a dense model.

Re: Qwen3-Next

#177

Post by jug »

dragonwriter wrote: post #157
jug wrote: post #156

> The Qwen3-Next-80B-A3B-Instruct performs comparably to our flagship model Qwen3-235B-A22B-Instruct-2507 I'm skeptical about these claims. How can this be? Wouldn't there be massive loss of world knowledge? I'm particularly skeptical because a recent trend in Q2 2025 has been benchmaxxing.

> I'm skeptical about these claims. How can this be? More efficient architecture. > Wouldn't there be massive loss of world knowledge? If you assume equally efficient architecture and no other salient differences, yes, that’s what you’d expect from a smaller model.

Hmm. Let's just say if this is true, that this is actually better with such a much lower total parameter count, it's the greatest accomplishment in over a year of LLM development. With the backdrop of bechmaxxing in 2025, I'll believe in this when I see the results on closed benchmarks and SimpleBench. My concern is this might be a hallucination machine.

Re: Qwen3-Next

#178

Post by bigyabai »

jug wrote: post #177
dragonwriter wrote: post #157

Earlier quoted context omitted.

> I'm skeptical about these claims. How can this be? More efficient architecture. > Wouldn't there be massive loss of world knowledge? If you assume equally efficient architecture and no other salient differences, yes, that’s what you’d expect from a smaller model.

Hmm. Let's just say if this is true, that this is actually better with such a much lower total parameter count, it's the greatest accomplishment in over a year of LLM development. With the backdrop of bechmaxxing in 2025, I'll believe in this when I see the results on closed benchmarks and SimpleBench. My concern is this might be a hallucination machine.

Might be. FWIW, my experience with the Qwen3 30b model basically took ChatGPT out of rotation for me. It's not hard for me to imagine an 80b model pushing that further, especially with thinking enabled.

I recommend playing with the free hosted models to draw your own conclusions: https://chat.qwen.ai/

Re: Qwen3-Next

#179

Post by takinola »

mmmllm wrote: post #82
amelius wrote: post #70

Earlier quoted context omitted.

https://en.wikipedia.org/wiki/Jevons_paradox

Sure but where is the demand going to come from? LLMs are already in every google search, in Whatsapp/Messenger, throughout Google workspace, Notion, Slack, etc. ChatGPT already has a billion users. Plus penetration is already very high in the areas where they are objectively useful: programming, customer care etc. I just don't see where the 100-1000x demand comes from to offset this. Would be happy to hear other vie…

I mean 640KB should be enough for anyone too but here we are. Assuming LLMs fulfill the expected vision, they will be in everything and everywhere. Think about how much the internet has permeated everyday life. Even my freaking toothbrush has WiFi now! 1000x demand is likely several orders of magnitude too low in terms of the potential demand (again, assuming LLMs deliver on the promise).

Re: Qwen3-Next

#180

Post by lanza »

mmmllm wrote: post #82
amelius wrote: post #70

Earlier quoted context omitted.

https://en.wikipedia.org/wiki/Jevons_paradox

Sure but where is the demand going to come from? LLMs are already in every google search, in Whatsapp/Messenger, throughout Google workspace, Notion, Slack, etc. ChatGPT already has a billion users. Plus penetration is already very high in the areas where they are objectively useful: programming, customer care etc. I just don't see where the 100-1000x demand comes from to offset this. Would be happy to hear other vie…

The difference in usefulness between ChatGPT free and ChatGPT Pro is significant. Turning up compute for each embedded usage of LLM inference will be a valid path forward for years.
Post reply on HN