38 comments

  • lnenad 26 minutes ago

    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.

    • grim_io 23 minutes ago

      That's low reasoning for a model, but max for a HN comment.

    • cyanydeez 13 minutes ago

      My stack is basically deer-flow with Qwen3.5-122B-A10B; this hopefully will be a speed and intelligence improvement. Running deer-flow overnight on any research topic or verify clear scoped programming issue is really neat.

      Also, heating my home during the winter is nice.

      Oh, also, I use llamacpp with --reasoning-budget; very simple way to move on.

  • rohansood15 42 minutes ago

    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.

    • user43928 31 minutes ago

      For comparison with hosted models, GPT 5.6 Luna scores 67% on DeepSWE, compared to 59% here for Qwen.

      Luna is $0.20 / $1.20 vs $0.16 / $0.47 with Qwen.

      • rohansood15 19 minutes ago

        This is a good counter argument. But you have to note that this is after OpenAI cut Luna costs by 80%. If you compare launch pricing, Qwen probably comes out ahead on a cost-performance basis.

        • jrflo 3 minutes ago

          The luna cost cuts were real though, not a one time promotion or something, due to some optimization (probably distillation?) that openai did.

    • Squarex 35 minutes ago

      I don't like these comparisons. Sure it is impressive, but it does not have a world knowledge of larger models. It has most of theirs intelligence.

      • rohansood15 22 minutes ago

        For world knowledge, you'd want it to find and reference the source material to be sure. At that point, it doesn't matter if the knowledge is embedded.

      • LaurensBER 26 minutes ago

        If/when we can get larger context this will mostly be mitigated by these smaller models being able to search the internet.

        Self-learning/improving would be even better but that's still a long way to go.

    • gruez 32 minutes ago

      >Opus 4.6 Max self-hosted at 30 tok/s on a 5k Macbook in Aug 2026. The LLM timelines are crazy.

      How much memory does this translate to and what quantization (if any) were applied?

  • pram 25 minutes ago

    It's in Unsloth Desktop already. Looks like it's 73GB, so 128GB Mac or Strix Halo etc will work. Exciting!

    • cwizou 20 minutes ago

      Download is available, but likely need to wait for an update, I get this which is understandable with the architectural change :

      Original error: llama.cpp does not support this GGUF's model architecture ('qwen4exp')

      Edit : Saw the pull request, should arrive soon enough https://github.com/ggml-org/llama.cpp/pull/27742

  • a_humean 11 minutes ago

    Waiting for llama.cpp support to land, but this might be a big deal for Strix Halo users.

    6B active params helps around the memory bandwidth constraints, but a 128GB box can probably run the Q3/Q4 quants fairly easily with a decent context size. This might actually be better for strix users than 27B, which was already very good.

  • tosh 42 minutes ago

    this is a new architecture (foreshadowing qwen 4)

    > trained at just 1/9 the cost of Qwen3.7-Plus, while outperforming it across the board

    https://x.com/Alibaba_Qwen/status/2092591393424515114

  • Imustaskforhelp a minute ago

    Pelican: https://gist.github.com/SerJaimeLannister/8fdef9c00175da0ca6...

    Aside from the pelican, I am sort of impressed by the fact that things are going the way in terms of really impressive small models.

    Also I love how this uses N-gram embedding. I think that Longcat was the first one who used it (I submitted that submission on hackernews because I really just loved the idea of it that I understood), I am certainly more interested in local LLM models and its interesting how they are utilizing new architectures to do some really impressive optimizations!

    (Do note that I created it using a free rate limited end-point that I found on the huggingface space section: victor-chat-with-qwen3-8-flash-next.hf.space)

  • armcat 16 minutes ago

    How is input token efficiency/verbosity on this model? Has anyone tried? GLM 5.2 was doing lot of turns and thinking piling up input tokens in the context (compared to Claude and GPT models). Then Qwen3.8-27B was 2x of that. Both delivered good output results but those cumulative input token costs were not cheap. Note this is on our specific business workloads. Genuinely interested in other people's experience (if you are able to try it out).

    • petu 6 minutes ago

      Haven't tried, would be surprised if it's any different.

      It's new arch demo for future Qwen 4 family, but (as I understand) training recipe/data is same as any other 3.8 model.

  • freakynit an hour ago

    Those benchmarks look seriously impressive.. considering how small of a MoE model this is.

  • martinald 26 minutes ago

    FYI: nothing seems to be able to run this (easily) yet. llama.cpp, vllm etc I couldn't get working because of no support in the mainline version.

  • amclennon 26 minutes ago

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

    • geooff_ 16 minutes ago

      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)

    • kzrdude 18 minutes ago

      Great. It's also smaller than DSV4 Flash, so it makes sense that way.

  • whwhyb an hour ago

    looks like it's better than deepseek v4 flash

  • KolmogorovComp 21 minutes ago

    Will this be cheaper than DS4flash ?

  • christkv 43 minutes ago

    Looks like a good model for strix halo

    • Iolaum 32 minutes ago

      indeed, can't wait for it to be supported by llama.cpp (or other engines)?

  • skarz an hour ago

    do we really need breaking news about qwen posted every single day?

    • KronisLV 34 minutes ago

      If there’s news, then yes. This is a pretty great new release for those still stuck on Qwen3.6 35B A3B if they have enough memory but don’t have super powerful compute.

      I wonder if I could get this running through vLLM on 6x Nvidia L4 - the 3.6 worked great on 4 cards but sadly TP6 just isn’t a thing and I don’t have 8 cards available, maybe it’s gonna be okay with like TP2 and MTP. I have no idea at this time, probably need to test out what even might be possible.

    • NitpickLawyer 30 minutes ago

      This particular release is interesting because it's a preview of qwen4 architecture. And, while benchmarks are iffy, this is a direct comparison, by the same team, with qwen3.8-27b that was pretty well received for a local model.

      This "next" release adds a new concept, first public release with n-grams, I think. And it's in a MoE size that is likely to be very fast and cheap to serve (faster than 27b for sure). It's also well suited for inference on alternative compute (i.e. sparks, macs, etc) so it's relevant to local users.

    • pseudony 37 minutes ago

      I and presumably quite a few others with AMD AI or Apple Mac platforms are very impacted by this.

      :)

      It is very relevant and for a certain group of us, far more impactful to our work the next month(s) than any blog post could be.

    • tosh 43 minutes ago

      this is a new architecture (foreshadowing qwen 4)

      > trained at just 1/9 the cost of Qwen3.7-Plus, while outperforming it across the board

      https://x.com/Alibaba_Qwen/status/2092591393424515114

    • c16 26 minutes ago

      There are many topics, personalities and politicians we hear about daily who have no merit.

      Qwen's advances do (currently) have merit.

    • dofm 27 minutes ago

      This actually is meaningful news, I think. Pretty wide audience appeal in the local LLM space too.

    • iAMkenough an hour ago

      yes there’s no shortage of online real estate