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

Qwen3.8-Flash-Next

qwen.ai

Post reply on HN
Print view View original

71–80 of 246 posts

Re: Qwen3.8-Flash-Next

#71

Post by kaelwd »

twohaibei wrote: post #40
geooff_ wrote: post #34

Earlier quoted context omitted.

Where are you seeing that? At the bottom of this post from Qwen I see: Qwen 3.8 flash: $0.16 / $0.47 Compared to Deepseek 0723: $0.03 / $0.075 (units in USD/m tok)

0.03 / 0.075 ? Where can i get that prices? Especially during peak hours DS4flash became much more money hungry than last month. https://api-docs.deepseek.com/quick_start/pricing

https://openrouter.ai/deepseek/deepseek-v4-flash-0731?endpoi...

8t/s though apparently and their cache hit rate is terrible so I don't think it's worth it over Relace.

Re: Qwen3.8-Flash-Next

#73

Post by lucabytheway »

very interesting. new architectures is the most interesting type of news. after what i experienced when gpt-oss came out i have been on the look out for architectural approaches that improves efficiency.

Re: Qwen3.8-Flash-Next

#74

Post by esafak »

lnenad wrote: post #21

Adding to my homelab stack, hopefully it doesn't overthink like the little model. Actually, hoping it thinks a bit less. Wait actually I'm really praying it reasons a bit more directly. But wait, I'm really sure that it must be a bit better.

It will be interesting to see the token efficiency analysis. This is my first question now with Chinese models; I take raw benchmark performance for granted.

Re: Qwen3.8-Flash-Next

#75

Post by dalant979 »

kzrdude wrote: post #32
amclennon wrote: post #24

It looks like this also undercuts the already absurdly inexpensive Deepseek Flash in pricing. Wild.

(Edited: I thought Qwen3.8 Flash Next was smaller, but it's not, in bytes. Here's how they compare.) DSV4 Flash 304B params, 167 GB download (at full size) Qwen3.8 Flash Next 180B params, 360 GB download (at full size)

180B?

Re: Qwen3.8-Flash-Next

#76

Post by Phemist »

khalic wrote: post #67
andy99 wrote: post #58

> Qwen3.8-Flash-Next features a 125B-parameter main model, supplemented by an additional 51B N-gram embeddings, with 6B parameters activated per token. Didn’t see this mentioned yet. I wonder what this means for the effective size. It’s evidently ~176B paramètres, but how does that get quantized. A 4-bit quant under 100GB seems unlikely, I’m suspecting this won’t run in 128GB unified memory In principle I like the id…

Gonna have to wait a few days to see what the wizards of the HF community come up with…

They are already working on it.

https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF https://unsloth.ai/docs/models/qwen3.8-next

> You will need at least 75 GB of RAM or unified memory to run the model. Its smallest 1-bit quantized version is larger than usual because of the model’s architecture so 1-bit isn't really 1-bit at all. However, this also means the quantization is less aggressive, allowing the model to retain more of its original accuracy than more heavily quantized models.

Lots of RAM required even for the 1-bit, which is already downloadable. Interested to see how well this one works compared to Ornith1.5-35B-A3B I've been running (and quite happy about).

Edit: but llama-cpp does not yet support it.

Re: Qwen3.8-Flash-Next

#77

Post by aftbit »

hedora wrote: post #69
rohansood15 wrote: post #9

Didn't expect it to beat 3.8 27B so cleanly. Opus 4.6 Max self-hosted at 30 tok/s on a 5k Macbook in Aug 2026. The LLM timelines are crazy.

My AMD strix halo box (haven’t benchmarked yet) should also run it reasonably well. It was $1400 at launch, and is $4K now. Your mac is < $2K in Biden-era dollars. Presumably the economy will eventually recover; maybe in one Moore’s law doubling if the midterms go outrageously well. That’ll be two doublings since the halo launched. I’d expect this model to run on a sub $1K box by then. $2K ought to get you a 512b par…

I believe you're underestimating the lag inherent in the economy. Even if we grant the idea that the political party controlling the US House/Senate has a significant impact on the economy, and that the current party is BAD and the next one would be GOOD, I would still expect that things will continue getting WORSE for a good 4 to 8 years before they get better again.

And that's even with assuming that we can continue to ignore the long-term problems like social security insolvency, the debt bomb, or climate change forever.

Re: Qwen3.8-Flash-Next

#78

Post by NewJazz »

hedora wrote: post #69
rohansood15 wrote: post #9

Didn't expect it to beat 3.8 27B so cleanly. Opus 4.6 Max self-hosted at 30 tok/s on a 5k Macbook in Aug 2026. The LLM timelines are crazy.

My AMD strix halo box (haven’t benchmarked yet) should also run it reasonably well. It was $1400 at launch, and is $4K now. Your mac is < $2K in Biden-era dollars. Presumably the economy will eventually recover; maybe in one Moore’s law doubling if the midterms go outrageously well. That’ll be two doublings since the halo launched. I’d expect this model to run on a sub $1K box by then. $2K ought to get you a 512b par…

You know the memory cartel isn't even close to being broken, right?

Re: Qwen3.8-Flash-Next

#79

Post by NitpickLawyer »

andy99 wrote: post #58

> Qwen3.8-Flash-Next features a 125B-parameter main model, supplemented by an additional 51B N-gram embeddings, with 6B parameters activated per token. Didn’t see this mentioned yet. I wonder what this means for the effective size. It’s evidently ~176B paramètres, but how does that get quantized. A 4-bit quant under 100GB seems unlikely, I’m suspecting this won’t run in 128GB unified memory In principle I like the id…

It is 125B A6B. vLLM is already out with support, ngrams can be offloaded to RAM so you only need ~96GB VRAM for nvfp4 w/ full context.

Likely soon we'll see nvme offloading for ngrams as well. They're just an index, so that should be plenty fast for what it does. LLama.cpp support should come soon as well, and they might do some things with offloading first.

Re: Qwen3.8-Flash-Next

#80

Post by raducu »

redrix wrote: post #53
lnenad wrote: post #21

Adding to my homelab stack, hopefully it doesn't overthink like the little model. Actually, hoping it thinks a bit less. Wait actually I'm really praying it reasons a bit more directly. But wait, I'm really sure that it must be a bit better.

You’re absolutely right to be hopeful. Three honest possibilities, and I’ll be straight with you about each: 1. It overthinks — Just like the previous iteration. High confidence. 2. It doesn’t overthink — Improvement from the last model for your use case. Regression for others. 3. It sometimes overthinks — Best case all around. A feature, not an impairment. One final thing worth mentioning: (I made myself irrationall…

On one hand I love your joke, on the other, this is HN not reddit and I usually downvote such responses, not sure what is the HN etiquette for such humor?
Post reply on HN