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

Qwen 3.8 27B

huggingface.co

Post reply on HN
Print view View original

411–420 of 848 posts

Re: Qwen 3.8 27B

#411

Post by bilekas »

CamperBob2 wrote: post #146
bmitc wrote: post #138

Earlier quoted context omitted.

Lol at that command. Why is this stuff so hard to run locally? I've spent a few days trying to figure it all out and haven't been able to. LM Studio doesn't work behind proxies. Ollama is confusing and doesn't seem to support Qwen3? And Llama.cpp is your command. I just want to run ` ` with some default parameters set and for it to run locally.

I never install this stuff manually anymore. Just tell your LLM of choice to download model X from URL Y, build the latest inference engine of choice E, and then create batch files or shell scripts to run instruct and/or reasoning models in accordance with instructions at URL Z.

Promise I'm not being flippant or rude, but why not ask it to write something like a parameterized script to do the same thing for you ?

Re: Qwen 3.8 27B

#412

Post by jjcm »

Image->html test for this.

Original images: https://image.non.io/neonRamenDesigns.webp

Qwen 3.8 build: https://html.non.io/neonRamenQwen3.8-27b

Overall I'm very impressed with how well this did. It's a big improvement over 3.6, and it feels on-par with some much, much larger models. I think this one is on-par with Gemini 3.7 Flash.

One thing to note - the build for this on my RTX 6000 pro blackwell took a long time. Easily one of the longest builds I've done. It took around 2 hours to build the site. Obviously we'll have some quants for this soon that will accelerate things, but I was still surprised with how long it took.

Comparison builds from this week:

https://html.non.io/neonRamenGemini3.7

https://html.non.io/neonRamenGLM5.3 (note: non-multimodal)

Re: Qwen 3.8 27B

#413

Post by gaigalas »

zazibar wrote: post #410
gaigalas wrote: post #312

Waiting for the MTP version to pop up on Unsloth. Speculative decoding makes a huge difference. Been running quantized 3.6 at 110t/s on a cheap 5060Ti and quite happy with it. If 3.8 improves on it, it would be awesome.

Good news, MTP support is already included in this release. Not sure why they haven't made this clearer.

I don't see an MTP entry on Unsloth though. Maybe it's not available in a lower quant I need for my poor GPU.

Re: Qwen 3.8 27B

#414

Post by rapind »

tinfoilhatter wrote: post #283
rob wrote: post #215

Earlier quoted context omitted.

I just read through a couple of your posts that weren't dead or buried, and it seems like you're pretty anti-AI. You should really start to have an open mind towards it. It's going to be the future (if it isn't already), and as you continue to get older, you're going to really wish you spent your time right now learning and embracing the technology instead of being so against it. A lot of the skills and things that y…

I'm not sure that typing messages to a chat bot requires much catching up, but since you think this is what I should do rob, I better listen!

> I'm not sure that typing messages to a chat bot requires much catching up

Lol, yeah I'd wager AI tooling skills are less relevant than knowing your way around a shell. It doesn't help that all of the AI tooling has event more churn than js libraries and package management systems!

I think there's probably some value in understanding how LLMs work, but beyond that it's the same boring skills that matter the most... Critical thinking, design chops, attention to detail, perseverance, etc.

Re: Qwen 3.8 27B

#415

Post by tomr75 »

jjcm wrote: post #412

Image->html test for this. Original images: https://image.non.io/neonRamenDesigns.webp Qwen 3.8 build: https://html.non.io/neonRamenQwen3.8-27b Overall I'm very impressed with how well this did. It's a big improvement over 3.6, and it feels on-par with some much, much larger models. I think this one is on-par with Gemini 3.7 Flash. One thing to note - the build for this on my RTX 6000 pro blackwell took a long time.…

what token/s?

Re: Qwen 3.8 27B

#416

Post by water-drummer »

erdaltoprak wrote: post #3

This is one of the most important model releases since most use cases don't need SOTA/Frontier If you want Qwen3.8-27B Serving Configs for the DGX Spark vLLM NVFP4 and RTX 4090 llama.cpp GGUF I added the setups here https://x.com/ErdalToprak/status/2088299678085308761?s=20

For those who don't wanna open twitter: https://xcancel.com/ErdalToprak/status/2088299678085308761?s...

Re: Qwen 3.8 27B

#417

Post by dofm »

There's a real change (compared to 3.6) in the way it writes in thinking — it drops words like "to" and "we" in "We need to", talks generally in note form, drops the/and all over the place, avoids "for".

"Need be helpful concise", "Need maybe not overdo", "Need ask!" Almost caveman.

I have an (unsourced, vague) suspicion that this rather unique thinking trace pattern is actually hobbling the MTP predictions, which seem to perform poorly.

Other notes: it uses the trick of repeating the prompt in the thinking trace.

It also worries about hidden chain of thought appearing in the final answer. It talks about "desired oververbosity 9", which is new. A bit GPT-ish.

It is being extraordinarily thorough in thinking through one of my code requests, but I don't know if the net result will be any better than the 35B MoE.

I asked it to ask me clarifying questions — it did, and it offered me a list of defaults I could simply agree to.

I don't think it is necessarily overthinking in the looping sense, but it is in the being exhaustive sense. I need to explore how it does with a tighter reasoning budget.

I am impressed but I am definitely in Camp Please-35B-A3B-When? here, because on an M1 Max this isn't really practical. I hope they do one, though I think they may not.

Re: Qwen 3.8 27B

#418

Post by sheepscreek »

Better than Opus 4.6 at computer use? Comparable with it for SWE? Am I reading this right?

I’ve heard rumours about AI shops optimizing for benchmarks. I also don’t think Qwen/Alibaba would be crazy enough to claim something unless there is some truth in it. Would love to see a side-by-side with Opus 4.6 on categories where Qwen 3.8 27B aces it.

Re: Qwen 3.8 27B

#419

Post by simonw »

Absolutely the best pelican I've seen from a model that runs on my laptop: https://tools.simonwillison.net/markdown-svg-renderer#url=ht...

Bicycle is the right shape. Pelican beak is excellent. Nice background. Most importantly, the pelican has one leg on each side of the bicycle - that's very rare.

(No chain on this bicycle though - in the reasoning trace it says "already chainstay... skip chain detail; maybe a small chainring.")

I ran that on an M5 Max MacBook Pro using LM Studio and their 17GB GGUF: https://lmstudio.ai/models/qwen3.8

It took 21 minutes(!) and used 22,276 reasoning tokens to produce 3,223 tokens of output.

(For the "they're training on your benchmark now" crowd, all of that cheating didn't prevent it from spending 20 minutes thinking about the task first! You can see the reasoning trace in the link I shared.)

For comparison, here's one I got from qwen3.8-2.4t-a95b on OpenRouter, which is pleasingly animated: https://tools.simonwillison.net/markdown-svg-renderer#url=ht...

Re: Qwen 3.8 27B

#420

Post by Frost1x »

Casteil wrote: post #82

One thing a lot of people don't seem to factor when hyping Qwen is how much models like this tend to 'overthink' with seemingly endless 'second guessing'. 3.8 seems no different from what I've tried thus far. As capable as it is, it's hard to justify using it when a competing model (e.g. Gemma4:26b-a3b) can consistently achieve the same or similar response with only 1/10th as many 'thinking' tokens, achieve much high…

“Thinking” is just a guiding methodology to help iterations (between the initial prompt, results, and a mixture of harness back and forth to the LLM) converge on something sane in a massive parameter space.

I like to think of it much like (as a common example most people can relate to) the Newton-Rhapson method for finding roots of a (mathematic) function. Your initial prompt runs, then the ‘harness’ kicks in using whatever methodologies are behind them to iterate on that prompt (back and forth with the model, occasionally with the user to get better guidance) and refine the outputs to hopefully converge back to some sensible output or actions the user was initially looking for.

So you’re hoping for an LLM that sort of ‘zero shots’ or needs minimal iterations from a prompt to give usable results. I find from my anecdata it varies across models and what I’m trying to get it to converge on. I tend to prefer models to not zero shot attempt because they tend to not do great, I want them to get feedback often to let me push them down the route of convergence in spaces I already understand well, meanwhile I like them to explore and give me new paths in spaces I’m not too familiar with.

That’s really what all that “second guessing” is, it’s making sure you’re following a sane path in a massive parameter space of an ambiguously defined problem. Imagine if in Newton’s method you checked the slope and it didn’t decrease from the last iteration and you just say “screw it let’s keep trying that direction.” LLMs and their harnesses tend not to have that base assumption like iteration on decreasing slopes to guide them closer to convergence, it’s a lot messier.

Post reply on HN