268
If open weight models are the future, U.S. AI companies are going to have a hard time (www.fastcompany.com)
submitted 3 months ago by to c/technology@lemmy.world
86 comments fedilink hide all child comments
you are viewing a single comment's thread
view the rest of the comments
[–] 22 points 3 months ago (20 children)

And Llama and Mistral are ancient history at this point.

The cutting edge of local is lightyears better now. It's basically where ChatGPT/Anthropic were not that long ago, with a bit less world knowledge because of the size.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 20 child comments
    •
  • [–] 10 points 3 months ago (16 children)

    What's the cutting edge now? Skool me...I want to try it. Can I grab one using ollama?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 16 child comments
    •
  • [–] 16 points 3 months ago* (14 children)

    https://sleepingrobots.com/dreams/stop-using-ollama/

    And this is just the tip of the iceberg for ollama. They're the same kind of scammy tech bros as OpenAI.

    The best setup depends on your hardware. There is no "easy button" unfortunately, quantized LLMs are just too intense and finicky to run without making some informed choices.

    It also depends on what you want to do with the LLM. For example, some are too slow or bad at long context for agenic use, some quantizations are great at scripts but terrible outside that, or vice versa.

    But LM Studio and Qwen 3.5 35B Q4 is probably the "easiest" flat recommendation I can make.

    Or... honestly, just pay $40 for basically unlimited usage for a year from an API, then roll your own frontend.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 14 child comments
    •
  • [–] 5 points 3 months ago (3 children)

    Or... honestly, just pay $40 for basically unlimited usage for a year from an API, then roll your own frontend.

    Can you clarify, what you mean by this? Rent a VPS? Or is there a legitimately good place that'll offer "unlimited" llm api access for $40 / year (and would you have any sort of privacy with this)?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 2 points 3 months ago (2 children)

    I was thinking of Xiaomi's Lite coding plan:

    https://platform.xiaomimimo.com/token-plan

    My mistake, it's $60. But that's still dirt cheap.

    You can read their terms of service; coding APIs are generally less invasive than something like the ChatGPT app, but don't bet on any privacy either.

    For my personal setup, I actually run most queries locally, via MiMo 2.5, but send "bulk" privacy-insensetive prompts to the same model on the API, since its so much faster. I kinda use them in parallel.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 1 point 3 months ago (1 child)

    Thanks for sharing the link. I'll take a peek at them and their ToS.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 1 point 3 months ago*

    I also forgot to emphasize this, but Xiaomi’s plan, in my opinion, is undiscovered fruit.

    GLM had a similar coding plan, but once it got in the news and popular, it got WAY more expensive and limited. I’m grandfathered into 6 more months of a GLM plan you literally cannot buy now.

    And I think Xiaomi is in the same situation GLM was 6+ months ago. It’s a fantastic model series, but unlike Kimi/GLM no one knows about it yet, which is how it’s still $60 for a year.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 3 points 3 months ago (1 child)

    Why are quantised LLMs harder to run than non quantised ones?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 9 points 3 months ago* (last edited 3 months ago)

    I just meant that you have to be cognizant of what went into the quantization.

    As an example, a “Q4_K_M” could be too much quantization to be usable on one model, and an inefficient waste of space on the other. Two Q4_K_Ms of the exact same model could be completely different, one totally borked. Or one particular Q4_K_M could excel in one task, but be totally useless for another, even with the exact same settings, when a slightly different sized or type of quantization would excel.

    It’s a deep rabbit hole. It’s not random either; there are distinct technical reasons behind every case mentioned above.

    And that’s not even at the cutting edge quantization anymore, though what’s “cutting edge” completely depends on your particular hardware and use case.

    I’m trying to make this sound daunting on purpose.

    Many people have really horrible experience with a default “ollama run” for this exact reason, because the defaults are terrible and the customization is critical to getting coherent, performant output.

    Unquantized LLMs, on the other hand, are basically always run the same way: vllm docker image on a big server, official weights. There’s less to “go wrong” trying to squeeze it on hardware with unofficial runtimes and compressors.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 2 points 3 months ago (7 children)

    Well that sucks. I was really impressed as a novice to open weight LLMs with the ease of use for Ollama on Bazzite.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 7 child comments
    •
  • [–] 4 points 3 months ago (6 children)

    I've been running LM Studio on Bazzite and I had to do nothing to get it working. Just go to the LM Studio website and download the .appimage for Linux. If you open it with Gear Lever it will install like an app from the app store and show up in your launcher with an icon.

    From there I have just been able to download models and use them from in the app. In fact I setup a local server to connect to my IDE and have been trying out local models for coding. It's pretty cool

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 6 child comments
    •
  • [–] 2 points 3 months ago (5 children)

    Awesome! I'll give it a try.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 5 child comments
    •
  • [–] 4 points 3 months ago (4 children)

    I can also vouch for lmstudio. If you can get Hermes running on Linux I would suggest trying that as well. It connects to lm studio and you use Hermes to communicate with the model. Iook into it as there's a lot to it, I've really been enjoying using it so far it even learns how I like to create tasks and I've stopped having to ask it to delegate certain tasks, it just knows to do it and to break down the tasks so my fairly context starved local model can handle it.

    As for a model, the Qwen 3.6 family of models do really well. I'd suggest the Qwen 3.6 35B a3b probably Q4 depending on your hardware. It's large, but because it's a mixture of experts model only 3b of experts are kept on vram at any one time so it stays fast. Qwen 3.6 27b is the smarter "dense" model, but trying to stay with Q4 for quality it becomes too large for 16GB vram and for me runs at like 2 tokens per second lol

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 2 points 3 months ago* (3 children)

    +1 for Hermes.

    If you have a newer Nvidia GPU, you can run Qwen 27B via exllamav3 and get good quality/speed in 16GB. And it's worth the trouble, as 27B is an amazing model.

    If it's AMD, yeah, a3B is a good bet, depending on how much spare CPU RAM you have.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 1 point 3 months ago (2 children)

    Thanks for the tip! I do have a 5080 so I'll have to look up exllamav3

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 2 points 3 months ago* (1 child)

    You want this one:

    https://huggingface.co/turboderp/Qwen3.6-27B-exl3_3.30bpw/tree/main

    Or maybe the 3.5bpw one if you don’t mind less context, or 3bpw if you need more:

    https://huggingface.co/turboderp/Qwen3.6-27B-exl3

    For faster inference at the cost of a little more VRAM usage:

    https://huggingface.co/turboderp/Qwen3.6-27B-DFlash-exl3

    And you run those in:

    https://github.com/theroyallab/tabbyAPI

    And FYI, if you have 64GB of RAM or more, you might consider hybrid inference instead.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 3 points 3 months ago

    Depends on your RAM (main + GPU), but assuming 32GB total: Qwen 3.6 35B A3B for coding support, Gemma 4 26B for general stuff. The LM Studio app curates a list of recommended models that will run well in it and makes it easy to run them.

    Mind you, what I like most about local models is their limitations, because it turns out closed models have limitations of the same nature, just with quite a bit more runway; and becoming aware of those limitations is valuable.

  • permalink
  • fedilink
  • source
  • parent
  • [–] 2 points 3 months ago (2 children)

    Thanks for the info. I don’t follow it closely but afaik wasn’t Mistral the only western open weight model around?

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 4 points 3 months ago (1 child)

    Mistral is still around, but for local LLMs... they're kinda irrelevant, sadly. Their models have regressed. I think they're being choked by ambiguous EU law.

    If you're looking for "western" weights, I'd look at the Laguna series first:

    https://huggingface.co/poolside

    And ik_llama.cpp quantizations like this: https://huggingface.co/sigargv/Laguna-M.1-GGUF/tree/main

    There's a couple of other interesting startups, but TBH its hard to keep track of where they're from.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •