This is further illustration that raw model capability isn’t really the most interesting thing at this point. The model and the harness should be viewed as a single system. I expect that we’ll be moving towards neurosymbolic systems where LLMs act as a heuristic engine within the boundaries of a symbolic logic system.

Possibly, I haven’t looked at how easy it is to get your hands on one of those.

If you absolutely need one now, a mac is probably the cheapest way to run them because of the unified memory. With any x86 solution you have to get a separate video card with at least 32gb vram to run a decent local model. However, if you wait a bit then you can probably get a dedicated chip a lot cheaper in the near future https://wccftech.com/alibabas-tsmc-built-5nm-risc-v-chip-xuantie-c950-now-runs-qwen-3-8-27b-model-natively-unlocking-massive-vertical-integration-tailwinds/

I expect doing ASICs for models will work even if they keep improving. It’ll be like regular chips getting new versions. You buy a chip with a specific model etched into it, and if it does what you need great. Next year, a new version comes out. So, it’s actually a feature since it allows companies to keep selling new chips.

It does look like we are entering diminishing returns territory though. The biggest evidence for this is that Chinese companies have now basically caught up to Anthropic and OpenAI. If the progress at the frontier was still happening at the same rate, then the gap wouldn’t be closing so quickly. There’s also a lot less noticeable difference between stuff like Claude 4.6 and Claude 5. When they went from 3.x to 4.x it was very noticeable. And at least for agentic coding, most of the improvement seems to come from the harness now. I expect improvements will continue, but at a much more gradual pace. It’s also possible people will figure out a new architecture that’s superior to LLMs, or works with them. World models are one promising area already being explored.

пошел на хуй либерал

That’s kind of what I’m expecting going forward too. Local models will get good enough in a year or two for most tasks, and then you just have a specialized chip like the GPU to run them.

Alibaba just announced a chip specifically for running local models. We’ll see what it ends up going for. https://wccftech.com/alibabas-tsmc-built-5nm-risc-v-chip-xuantie-c950-now-runs-qwen-3-8-27b-model-natively-unlocking-massive-vertical-integration-tailwinds/

Training happens once per model, but inference is an ongoing process. So, there’s going to be a huge amount of energy saving if we move to using local models.

Thank you for taking your time away from sniffing glue to share your invaluable insights.

Right, Wall Street Journal and other western media cited are doing propaganda for DPRK. That’s definitely an interesting take you’ll have to expand on.

enjoy being educated https://dialecticaldispatches.substack.com/p/the-wsj-just-confirmed-what-socialists

a lost redditor appears, it is very confused

haha I keep forgetting how insane prices are now

How’s the output quality on the lowest quant?

I think it’ll depend on how much these chips cost. If it’s gonna be something like a Raspberry Pi that works as well as a 500 dollar GPU, that’s gonna be pretty amazing.

a lost redditor appears, it is very confused

lmfao yeah

I’m not aware of the phenomenon outside the west, it’s kind of like anarchism in that way.

nice

Context matters, when it comes to objective facts, a bunch of people thinking something is true doesn’t make it so. However, when it comes to subjective experience, then the opinions of people actually living in the country do matter far more than those of the outsiders.