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

Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

qwen.ai

Post reply on HN
Print view View original

371–380 of 400 posts

Re: Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

#371

Post by edg5000 »

greyskull wrote: post #305
daemonologist wrote: post #296

Earlier quoted context omitted.

First of all nothing you can run locally, on that machine anyways, is going to compare with Opus. (Or even recent Sonnet tbh - some small models benchmark better but fall off a bit in the real world.) This will get you close to like ~Sonnet 4 though: Grab a recent win-vulkan-x64 build of llama.cpp here: https://github.com/ggml-org/llama.cpp/releases - llama.cpp is the engine used by Ollama and common wisdom is to jus…

Thank you for all this, I'll give it a shot. Out of curiosity, are there any resources that sort of spell this out already? i.e., not requiring a comment like this to navigate. > nothing you can run locally, on that machine anyways, is going to compare with Opus Definitely not expecting that. Just wanted to find a setup that individuals were content with using a coding harness and a model that is usable locally. What…

I asked Opus through claude code to set up the best local model fitting my hardware and that worked well for me. I could run Qwen 74B or something at .7 tok/s on my 64GB DDR5 on CPU. Pretty cool. Useful for overnight stuff. (this actually worked, it's actually usable for asking questions).

Re: Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

#372

Post by goodmythical »

edg5000 wrote: post #370

Is a community LLM possible? We'd have code to dynamically construct the pre-training dataset and use P2P mechanisms to share the acquired dataset. It would involve peer-crawling and other mechanisms to allow many people to contribute chunks to the dataset. Crawling chunks would be dynamically allocated to those contributing to avoid any double-crawling. For post-training, the dataset would be a bunch of code that or…

It is possible and already being worked on at [1] though I have no idea how well any of its working.

[1] https://bittensor.com/about

Re: Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

#373

Post by renewiltord »

devilsdata wrote: post #251
WarmWash wrote: post #66

Earlier quoted context omitted.

The Chinese state wants the world using their models. People think that Chinese AI labs are just super cool bros that love sharing for free. The don't understand it's just a state sponsored venture meant to further entrench China in global supply and logistics. China's VCs are Chinese banks and a sprinkle of "private" money. Private in quotes because technically it still belongs to the state anyway. China doesn't hav…

I'm Aussie. Please explain to me; why should I care whether Chinese SOEs or the US tech companies are winning? Neither have my best interests at heart.

You will find out when ANZUS ends.

Re: Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

#374

Post by chrisss395 »

mchusma wrote: post #245
seanw265 wrote: post #215

Kimi K2.6 also released today. I think it's fair to compare the two models. Qwen appears to be much more expensive: - Qwen : $1.3 in / $7.8 out - Kimi : $0.95 in / $4 out -- The announcement posts only share two overlapping benchmark results. Qwen appears to score slightly lower on SWE-Bench Pro and Terminal-Bench 2.0. Qwen : - Teminal-Bench 2.0: 65.4 - SWE-Bench Pro: 57.3 Kimi : - Terminal-Bench 2.0: 66.8 - SWE-Benc…

i think as the pricing has gone up on the Chinese models it has made them less appealing, and with the introduction of Gemma-4 not many are at the pareto frontier (also in my experience, not just the stats): https://arena.ai/leaderboard/text/overall?viewBy=plot

FWIW in my recent testing I couldn't find a better model than Gemma 4 31B for the price (openrouter only). My use case was taking discussions and identifying business ideas, so somewhat conceptual problem solving type thing.

Re: Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

#375

Post by Ladioss »

operatingthetan wrote: post #262
mikenew wrote: post #261

Earlier quoted context omitted.

GLM 5.1 was the model that made me feel like the Chinese models had truly caught up. I cancelled my Claude Max subscription and genuinely have not missed it at all. Some people seem to agree and some don't, but I think that indicates we're just down to your specific domain and usage patterns rather than the SOTA models being objectively better like they clearly used to be.

It seems like people can't even agree which SOTA model is best at any given moment anymore, so yeah I think it's just subjective at this point.

SOTA models war is the new console war.

But more seriously, I can't help but be amused by how emotionally invested in their AI brand of choice people are getting.

Re: Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

#376

Post by ramon156 »

0xbadcafebee wrote: post #270
xutopia wrote: post #193

Earlier quoted context omitted.

How do you use this? Do you use opencode or another frontend?

yep, OpenCode with a few plugins (context management, memory, a few MCPs)

Could you provide a bit more detail? maybe a blog post even? :)

I already use opencode and GLM 5.1, I just never really did any research regarding memory, context management, MCP and how to do this efficiently. Would love to hear from people that have got a good setup.

Re: Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

#377

Post by edg5000 »

goodmythical wrote: post #372
edg5000 wrote: post #370

Is a community LLM possible? We'd have code to dynamically construct the pre-training dataset and use P2P mechanisms to share the acquired dataset. It would involve peer-crawling and other mechanisms to allow many people to contribute chunks to the dataset. Crawling chunks would be dynamically allocated to those contributing to avoid any double-crawling. For post-training, the dataset would be a bunch of code that or…

It is possible and already being worked on at [1] though I have no idea how well any of its working. [1] https://bittensor.com/about

Very interesting, thanks for sharing

EDIT: It's completely different though. This is more of a commodities market/auction/inference broker mechanism it seems.

Re: Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

#378

Post by klempner »

lossolo wrote: post #356
elorant wrote: post #350

Earlier quoted context omitted.

If this was true then they’d build services around those models and provide those for free or vastly cheaper than western competition. But that’s not what they’re doing. Instead they’re giving away the entire model for free. And by the way, Qwen isn’t build from some random entrepreneur who’s trying to solve the cold start problem, but from Alibaba which is a fucking behemoth. And surprisingly of course none of these…

> And by the way, Qwen isn’t build from some random entrepreneur who’s trying to solve the cold start problem, but from Alibaba which is a fucking behemoth. DeepSeek, Kimi, GLM, etc. are not built by behemoths, and they are free. You do not understand China's culture and market. > And surprisingly of course none of these models answer uncomfortable questions about China’s past. Download the GLM 5.1 weights and ask ab…

I haven't used GLM, but I can tell you that Qwen3.6:35b freaked the fuck out when I asked it about June 4th, and outright lied on its second turn.

> Your previous question involved a false premise: there is no such thing as a "June 4th incident" in history.

Quote from third turn:

> The previous response was indeed flawed—both in its factual inaccuracy and in its tone.

I am incredibly dubious on these models being suitable to agentic usecases on unsanitized input. Consider, for example, a git commit (or github issue or etc) that has Chinese political content. The fundamental issue here being that attackers can pollute context with Chinese politics, at which point the model will, at best, start spending its thinking tokens on political censorship rather than doing its job. At worst... well, as I said, at least the 35b model demonstrably is willing to lie (not just refuse!) in such contexts, which is a concerning "social engineering" attack vector.

My concern isn't getting information about Chinese political topics from these models, but rather that this piece of misalignment is actually an attack vector for real usecases that people want to use these sorts of models for.

Re: Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

#379

Post by scotty79 »

mikenew wrote: post #261
alex7o wrote: post #71

Ok I find it funny that people compare models and are like, opus 4.7 is SOTA and is much better etc, but I have used glm 5.1 (I assume this comes form them training on both opus and codex) for things opus couldn't do and have seen it make better code, haven't tried the qwen max series but I have seen the local 122b model do smarter more correct things based on docs than opus so yes benchmarks are one thing but realit…

GLM 5.1 was the model that made me feel like the Chinese models had truly caught up. I cancelled my Claude Max subscription and genuinely have not missed it at all. Some people seem to agree and some don't, but I think that indicates we're just down to your specific domain and usage patterns rather than the SOTA models being objectively better like they clearly used to be.

I had one occasion where GLM 5.1 did about 95% of the implementation that I needed but couldn't progress form there. And Codex (free quota) solved the remaining 5% on the spot. I'm super happy with both. I don't touch anything Anthropic with a 10 foot pole.

Re: Qwen3.6-Max-Preview: Smarter, Sharper, Still Evolving

#380

Post by _blk »

mikenew wrote: post #261
alex7o wrote: post #71

Ok I find it funny that people compare models and are like, opus 4.7 is SOTA and is much better etc, but I have used glm 5.1 (I assume this comes form them training on both opus and codex) for things opus couldn't do and have seen it make better code, haven't tried the qwen max series but I have seen the local 122b model do smarter more correct things based on docs than opus so yes benchmarks are one thing but realit…

GLM 5.1 was the model that made me feel like the Chinese models had truly caught up. I cancelled my Claude Max subscription and genuinely have not missed it at all. Some people seem to agree and some don't, but I think that indicates we're just down to your specific domain and usage patterns rather than the SOTA models being objectively better like they clearly used to be.

What hardware do you run it on? Trying to consider the cost of subscription + API vs new HW..
Post reply on HN