Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers(venturebeat.com)
91 pointsby lostmsu2 hours ago12 comments
  • alexpotato21 minutes ago
    I recently wrote a guide on getting:

    - llama.cpp

    - OpenCode

    - Qwen3-Coder-30B-A3B-Instruct in GGUF format (Q4_K_M quantization)

    working on a M1 MacBook Pro (e.g. using brew).

    It was bit finicky to get all of the pieces together so hopefully this can be used with these newer models.

    https://gist.github.com/alexpotato/5b76989c24593962898294038...

    • copperx20 minutes ago
      How fast does it run on your M1?
  • solarkraft40 minutes ago
    Smells like hyperbole. A lot of people making such claims don’t seem to have continued real world experience with these models or seem to have very weird standards for what they consider usable.

    Up until relatively recently, while people had already long been making these claims, it came with the asterisks of „oh, but you can’t practically use more than a few K tokens of context“.

    • tempest_19 minutes ago
      Qwen3-Coder-30B-A3B-Instruct is good I think for in line IDE integration or operating on small functions or library code but I dont think you will get too far with one shot feature implementation that people are currently doing with Claude or whatever.
  • solarkraft43 minutes ago
    What are the recommended 4 bit quants for the 35B model? I don’t see official ones: https://huggingface.co/models?other=base_model:quantized:Qwe...

    Edit: The unsloth quants seem to have been fixed, so they are probably the go-to again: https://unsloth.ai/docs/models/qwen3.5/gguf-benchmarks

  • kristianpaul13 minutes ago
    https://unsloth.ai/docs/models/qwen3.5#qwen3.5-27b “ Qwen3.5-27B For this guide we will be utilizing Dynamic 4-bit which works great on a 18GB RAM”
    • kristianp4 minutes ago
      18GB was an odd 3-channel one-off for the M3 Pros. I guess there's a bunch of them out there, but how slow would 27B be on it, due to not being an MOE model.
  • sunkeeh13 minutes ago
    Qwen3.5-122B-A10B BF16 GGUF = 224GB. The "80Gb VRAM" mentioned here will barely fit Q4_K_S (70GB), which will NOT perform as shown on benchmarks.

    Quite misleading, really.

  • gunalx10 minutes ago
    qwen 3.5 is really decent. oOtside for some weird failures on some scaffolding with seemingly different trained tools.

    Strong vision and reasoning performance, and the 35-a3b model run s pretty ok on a 16gb GPU with some CPU layers.

  • mark_l_watsonan hour ago
    The new 35b model is great. That said, it has slight incompatibility's with Claude Code. It is very good for tool use.
  • erelongan hour ago
    What kind of hardware does HN recommend or like to run these models?
    • suprjami43 minutes ago
      The cheapest option is two 3060 12G cards. You'll be able to fit the Q4 of the 27B or 35B with an okay context window.

      If you want to spend twice as much for more speed, get a 3090/4090/5090.

      If you want long context, get two of them.

      If you have enough spare cash to buy a car, get an RTX Ada with 96G VRAM.

      • barrkel10 minutes ago
        Rtx 6000 pro Blackwell, not ada, for 96GB.
    • andsoitis24 minutes ago
      For fast inference, you’d be hard pressed to beat an Nvidia RTX 5090 GPU.

      Check out the HP Omen 45L Max: https://www.hp.com/us-en/shop/pdp/omen-max-45l-gaming-dt-gt2...

      • laweijfmvo8 minutes ago
        I never would have guessed that in 2026, data centers would be measured in Watts and desktop PCs measured in liters.
    • zozbot23420 minutes ago
      It depends. How much are you willing to wait for an answer? Also, how far are you willing to push quantization, given the risk of degraded answers at more extreme quantization levels?
    • dajonker32 minutes ago
      Radeon R9700 with 32 GB VRAM is relatively affordable for the amount of RAM and with llama.cpp it runs fast enough for most things. These are workstation cards with blower fans and they are LOUD. Otherwise if you have the money to burn get a 5090 for speeeed and relatively low noise, especially if you limit power usage.
    • elorant19 minutes ago
      Macs or a strix halo. Unless you want to go lower than 8-bit quantization where any GPU with 24GBs of VRAM would probably run it.
    • xienzean hour ago
      It's less than you'd think. I'm using the 35B-A3B model on an A5000, which is something like a slightly faster 3080 with 24GB VRAM. I'm able to fit the entire Q4 model in memory with 128K context (and I think I would probably be able to do 256K since I still have like 4GB of VRAM free). The prompt processing is something like 1K tokens/second and generates around 100 tokens/second. Plenty fast for agentic use via Opencode.
      • rahimnathwani39 minutes ago
        There seem to be a lot of different Q4s of this model: https://www.reddit.com/r/LocalLLaMA/s/kHUnFWZXom

        I'm curious which one you're using.

        • suprjami33 minutes ago
          Unsloth Dynamic. Don't bother with anything else.
      • msuniverse202631 minutes ago
        I've had an AMD card for the last 5 years, so I kinda just tuned out of local LLM releases because AMD seemed to abandon rocm for my card (6900xt) - Is AMD capable of anything these days?
        • wirybeige12 minutes ago
          The vulkan backend for llama.cpp isn't that far behind rocm for pp and tp speeds
    • CamperBob220 minutes ago
      I think the 27B dense model at full precision and 122B MoE at 4- or 6-bit quantization are legitimate killer apps for the 96 GB RTX 6000 Pro Blackwell, if the budget supports it.

      I imagine any 24 GB card can run the lower quants at a reasonable rate, though, and those are still very good models.

      Big fan of Qwen 3.5. It actually delivers on some of the hype that the previous wave of open models never lived up to.

      • MarsIronPI11 minutes ago
        I've had good experience with GLM-4.7 and GLM-5.0. How would you compare them with Qwen 3.5? (If you have any experience with them.)
  • kristianpaul22 minutes ago
    They work great with kagi and pi
  • aliljetan hour ago
    Is this actually true? I want to see actual evals that match this up with Sonnet 4.5.
    • magicalhippo17 minutes ago
      The Qwen3.5 27B model did almost the same as Sonnet 4.5 in this[1] reasoning benchmark, results here[2].

      Obviously there's more to a model than that but it's a data point.

      [1]: https://github.com/fairydreaming/lineage-bench

      [2]: https://github.com/fairydreaming/lineage-bench-results/tree/...

    • lostmsuan hour ago
      Not exactly, but pretty close: https://artificialanalysis.ai/models/capabilities/coding?mod...

      Somewhere between Haiku 4.5 and Sonnet 4.5

      • CharlesWan hour ago
        > Somewhere between Haiku 4.5 and Sonnet 4.5

        That's like saying "somewhere between Eliza and Haiku 4.5". Haiku is not even a so-called 'reasoning model'.¹

        ¹ To preempt the easily-offended, this is what the latest Opus 4.6 in today's Claude Code update says: "Claude Haiku 4.5 is not a reasoning model — it's optimized for speed and cost efficiency. It's the fastest model in the Claude family, good for quick, straightforward tasks, but it doesn't have extended thinking/reasoning capabilities."

        • pityJuke25 minutes ago
          Haiku 4.5 is a reasoning model. [0]

          [0]: https://www-cdn.anthropic.com/7aad69bf12627d42234e01ee7c3630...

          > Claude Haiku 4.5, a new hybrid reasoning large language model from Anthropic in our small, fast model class.

          > As with each model released by Anthropic beginning with Claude Sonnet 3.7, Claude Haiku 4.5 is a hybrid reasoning model. This means that by default the model will answer a query rapidly, but users have the option to toggle on “extended thinking mode”, where the model will spend more time considering its response before it answers. Note that our previous model in the Haiku small-model class, Claude Haiku 3.5, did not have an extended thinking mode.

          • CharlesW15 minutes ago
            Sure, marketing people gonna market. But Haiku's 'extended thinking' mode is very different than the reasoning capabilities of Sonnet or Opus.

            I would absolutely believe mar-ticles that Qwen has achieved Haiku 4.5 'extended thinking' levels of coding prowess.

            • DetroitThrow6 minutes ago
              >Sure, marketing people gonna market.

              Oh HN never change.

      • pinum34 minutes ago
        Looks much closer to Haiku than Sonnet.

        Maybe "Qwen3.5 122B offers Haiku 4.5 performance on local computers" would be a more realistic and defensible claim.

  • xenospn2 hours ago
    Are there any non-Chinese open models that offer comparable performance?
    • MarsIronPI9 minutes ago
      I think you could look into Minstral. There's also GPT-OSS but I'm not sure how well it stacks up.

      What's your problem with Chinese LLMs?

  • u1hcw9nx2 hours ago
    [flagged]