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

Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

qwen.ai

Post reply on HN
Print view View original

441–450 of 482 posts

Re: Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

#441

Post by proxysna »

NwtnsMthd wrote: post #416
proxysna wrote: post #31

Earlier quoted context omitted.

Qwen3.5-27B with a 4bit quant can be run on a 24G card with no problem. With 2 Nvidia L4 cards and some additional vllm flags, i am serving 10 developers at 20-25tok/sek, off-peak is around 40tok/sek. Developers are ok with that performance, but ofc they requested more GPU's for added throughput.

Just curious, what's your setup like? How do the devs interact with the model?

OpenWebUI with postgres and vllm for inference, searxng for websearch a few other mcp's for tools.

Re: Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

#442

Post by solenoid0937 »

swiftcoder wrote: post #398
solenoid0937 wrote: post #381

Earlier quoted context omitted.

"Literally never runs out of tokens?" lol, no. Tokens are just energy. There is always a way to run out of tokens, and no one will subsidize free tokens forever.

"Never runs out of tokens" in the sense that running 8 hours a day 7 days a week is still under the subscription limit

You can also do that on an API without hitting a limit!

Re: Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

#443

Post by user_7832 »

mudkipdev wrote: post #295
mft_ wrote: post #271

Huh, running the Q4_K_M quant with LM Studio, and asked it "How can I set up Qwen 3.6 27b to use tools and access the local file system?". Part of its reply was: Quick clarification: As of early 2025, "Qwen 3.6" hasn't been released yet. You are likely looking for Qwen2.5, specifically the Qwen2.5-32B-Instruct model, which is the 30B-class model closest to your 27B reference. The instructions below will use this mode…

If you are talking with Claude about AI, it will sometimes passively bring up "frontier models like GPT-4o"

Slightly tangential, how good/bad is 4o compared to the modern (5.3 I think?) one?

TBH I personally find non-thinking replies quite poor for the type of questions I ask so I haven't touched chatgpt for months (ever since Gemini 2.5 Pro I think.) (And even Gemini 3.1 Pro tends to still be too literal at times instead of understand the implied meaning lol. We've got more place to improve.)

Re: Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

#444

Post by hex4def6 »

nopelican wrote: post #419
simonw wrote: post #106

The pelican is excellent for a 16.8GB quantized local model: https://simonwillison.net/2026/Apr/22/qwen36-27b/ I ran it on an M5 Pro with 128GB of RAM, but it only needs ~20GB of that. I expect it will run OK on a 32GB machine. Performance numbers: Reading: 20 tokens, 0.4s, 54.32 tokens/s Generation: 4,444 tokens, 2min 53s, 25.57 tokens/s I like it better than the pelican I got from Opus 4.7 the other day: https://si…

I just create the nopelican user to avoid seeing the same type of comments for scoring new models. Why doesn't someone create a pelican by month thread, like who is hiring, so that all who want to talk about their prefered mode and pelican can post with leisure at full extend. Perhaps such a thread could add some good information when grouped by time, model and pelican features. But I, honestly, think that the pelica…

The point of a benchmark is that it allows a relative comparison. The Pelican is one such benchmark.

Feel free to create a "how does it compare to Claude 3.5 Sonnet" benchmark. If people find it useful, it will be run against new LLMs to generate additional points of comparison.

I will also say; it's really easy to just skim past comments. I suspect your ROI time-wise in creating this account to complain will never be recouped compared with just skimming past pelican comment chains.

Re: Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

#445

Post by andriy_koval »

ginko wrote: post #403
fnordpiglet wrote: post #84

Earlier quoted context omitted.

For coding often quality at the margin is crucial even at a premium. It’s not the same as cranking out spam emails or HN posts at scale. This is why the marginal difference between your median engineer and your P99 engineer is comp is substantial, while the marginal comp difference between your median pick and packer vs your P99 pick and packer isn’t. I’d also say it keeps the frontier shops competitive while costing…

>but if you’re not a Chinese shop, would you really host your production code development workload on a Chinese hosted provider? The point of open source models is that you host them locally. I trust neither Chinese nor American providers with this.

another point is that there could be multiple inference providers, so market will be healthier, and not dominated by one player who charges NN% margin.

Re: Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

#446

Post by andriy_koval »

Zetaphor wrote: post #358
DiogenesKynikos wrote: post #346

Earlier quoted context omitted.

Are you claiming that major Chinese cloud providers like Tencent and Alibaba are pilfering trade secrets from their customers' data? To my knowledge, there's no evidence for that whatsoever. If it were true and came out, it would instantly tank their cloud businesses (which is why they don't do it, and why AWS, Azure, etc. also don't do it). If it were to happen, Chinese law does offer recourse, including to foreign…

In every one of these threads for a new Chinese open weights model, it's always the same tired discussion of how this is all actually a psyop by the Chinese government to undermine US interests and it can't answer questions about Tienanmen Square. Meanwhile I'm over here solving real world business problems with a model that I can securely run on-prem and not pay out the nose for cloud GPU inference. And then after w…

> The last thread for Qwen's new hosted model was full of folks talking about how "China" is no longer releasing open weights models, when the next day Moonshot AI releases Kimi 2.6. A few days later and here's Qwen again with another open release.

looks like you declared win argument, because you now see that 2.6 was released, but at that time your opponents argument stand.

Also, you can't predict if Chinese labs will continue releasing open frontier models. Looks like Kimi is the only one left, Qwen is much smaller model.

Re: Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

#447

Post by hadlock »

anonym29 wrote: post #315
hadlock wrote: post #264

Earlier quoted context omitted.

How many t/s output are you getting at Q4_K_M with 200k context on your Strix Halo if you ask it to add a new feature to a codebase.

Qwen 3.6 27B, and other dense models, as opposed to MoE models do NOT scale well. Like I said in my original post, for 27B usage specifically, I'd take a dGPU with 32GB of VRAM over Strix Halo. I also don't usually benchmark out to 200k, my typical depths are 0, 16k, 32k, 64k, 128k. That said, with Qwen 3.5 122B A10B, I am still getting 70 tok/s PP speed and 20 tok/s TG speed at 128k depth, and with Nemotron 3 Super…

Thanks this is very helpful for planning out localLLM buy. Sounds like we are still at least 1 generation out (DDR6 500-700GB/s memory) from getting to that magic ~25-30TG/s. Nemotron 3 Super architecture sounds promising.

Re: Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

#448

Post by gverrilla »

simonw wrote: post #106

The pelican is excellent for a 16.8GB quantized local model: https://simonwillison.net/2026/Apr/22/qwen36-27b/ I ran it on an M5 Pro with 128GB of RAM, but it only needs ~20GB of that. I expect it will run OK on a 32GB machine. Performance numbers: Reading: 20 tokens, 0.4s, 54.32 tokens/s Generation: 4,444 tokens, 2min 53s, 25.57 tokens/s I like it better than the pelican I got from Opus 4.7 the other day: https://si…

Time for a spin, mate.

Re: Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

#449

Post by swiftcoder »

solenoid0937 wrote: post #442
swiftcoder wrote: post #398

Earlier quoted context omitted.

"Never runs out of tokens" in the sense that running 8 hours a day 7 days a week is still under the subscription limit

You can also do that on an API without hitting a limit!

Not typically at predictable monthly spend, which turns out to be important to some folks

Re: Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model

#450

Post by swiftcoder »

sumedh wrote: post #288
swiftcoder wrote: post #249

Earlier quoted context omitted.

> For coding often quality at the margin is crucial even at a premium For some problems, sure, and when you are stuck, throwing tokens at Opus is worthwhile. On the other hand, a $10/month minimax 2.7 coding subscription that literally never runs out of tokens will happily perform most day-to-day coding tasks

Why pay for two subscriptions though? Claude also has other models which use less tokens.

Redundancy, mostly. And having left over tokens when Opus eats all of those tokens
Post reply on HN