I've just started using qwen3.6:35b a couple days ago running on my framework desktop and rather impressed. It runs really well and reminds me of probably the first Claude model I used. It's the first local model that's actually working for me in a coding agent I've tried. Very exciting!
Interesting. I wonder if there is opportunity to train a set of small model variants to excel at a certain stacks. Eg Qwen3.6-27B for Node + React or Qwen3.6-27B for Rust + TUI
This is always how I've imagined small/consumer-hardware models going in time. If I only ever code in Python, give me a model that does just that (plus some general CS, algorithms, structure, etc.) and does it super-fast and well. Make it small enough that if I need a Python back end and an HTML front end, another specific model can load alongside and collaborate on the front end.
Or give me a pure shopping model that has a general understanding of products and product categories, and then will playwright/scrape/API into shopping sites to compare options and find me what I want. Etc.
Qwen 3.6 27B is an anomalously strong all-around model for its size, but when we run our evaluations, we generate 10 coding submissions/language/model (110 total). So full discosure, the per-language per-model performances can be noisy (I do not think Qwen3.6 27B is better than Fable 5 in agentic workflows when writing Kotlin, given enough samples, although we do find some interesting anomalies that hold up under large sample sizes).
It is, but it's way too slow on a Strix Halo due to its limited bandwidth.
(I'm still sad that they didn't make a 122B-A10B version of it, as it's the kind of model that fits best on a Strix Halo, and for 3.5 it was comparable in performance to the dense 27B version).
Yeah the speed is vastly different but it's getting ~10 tps. And the MoE model is like 50 or something.
Might use it if it proves to be much smarter I dont really monitor my agents while they're running.
Off the top of my head since it seems to be the quick info you're looking for: IIRC, with these two, the 27B is a dense model, meaning it's all active at inference. Meanwhile, the 35B is a Mixture of Experts (MoE), so only part of its network (3B?) is active at any time.
Thanks! Dense models have been slow on my compute, but I'll give it a try. If its not toooooo slow then it's fine I mostly fire and forget agents anyway.
Edit: seems fast! I'll try it out some more, thanks again.
I'm running qwen36.:35b:iq4 IQ4_XS quant. Takes 18 GB of RAM with 131k context window. Seems to be really good. Have it running local stuff via Hermes, using a cloud model via Ollama (Deepseek V4-Pro) for heavy lifting.
I settled on opencode after trying goose and aider as well. I'll probably try some more but opencode worked similar to Claude code which is my main agent.
I serve the model with ollama and am thinking about replacing ollama but haven't looked into it.
I have openwebui for chat if I want that too, but don't really use it.
One is regulatory. At least in the US, every nuclear reactor that produces at least 100 MW needs to carry a 375 million dollar insurance policy at minimum. Under 100 MW there is an alternate schedule that ranges from 5 million to 75 million scaling based on output. But the net result is that it's still more profitable to built a single large reactor, since a 1 GW reactor is less to insure than 10 100 MW reactors. This is written into law, it would require Congress to change it.
Second is that nuclear reactor efficiency tends to improve with size. The ratio of thermal watts to electric watts tends to be better with large reactors. I'm not super well versed on the engineering tradeoffs here by my rough understanding is that waste heat scales with surface area while useful energy extraction scales with volume.
In the grand scheme of things, that doesn't seem to be too expensive. According to the NRC, insurance is per site, and additional reactors on a site don't increase the insurance nearly as much as the first. The example they give is $1.1 million annual premium for a $500 million policy, with multi-reactor sites going up to $1.5 million. They also mention property insurance ($1.06 billion policy) and they didn't discuss the premiums on that, but as it's not liability insurance it's probably cheaper.
The big costs are still going to be the cost of siting and building the reactor, the fuel, and the ongoing cost of running it. They pay off over a very long time horizon, so it's also the opportunity cost.
Really all you need is a good cli. Agents I use have no trouble with gh or glab or tea. Clicking on this git platform idk what the hell is going on. I clicked on one repo with random letters in the title and it was empty. Womp.
Have you heard of the term "bad faith argument"? That's a good example.
> - Opposition to political and cultural liberalism
> Yes, banned speech on twitter that trans people use.
I had to check this one. I found two stories. [1] shows that people were warned about speech (e.g. using the term cisgender) but were still able to post without problems. [2] showed that X's rules were updated to prevent harassment in the form of deadnaming or misgendering. I doubt that either supports your assertion to be true.
As to the rest of your answers, you seem to have given away that you didn't read the Britannica article at all:
> - Education as character building
> what
You should know, right? You read the article and made the accusation. You've done this several times:
> - The leadership principle
> idk what this is
> - The “new man”
> or this
There's not really a need for me to be pedantic in the face of such slipshod work, but then again, perhaps you think that not being slipshod is being pedantic. Which would also be wrong, but telling.
Try learning about what fascism is, that would be my suggestion. With a book, while putting your phone away.
pretty much exactly what i expected, you ignored all the YESs and called me dumb and made a ton of assumptions. You dont have to fit a definition 100% you're going to hang on to all the tiny little details and say wholey that he's not a fascists. I think he hit's the major point of the definition. Pedantic bootlicker. The reason I didn't pick up a book at your request is because I'm not wasting my time with you. And, no I didn't read the article, you did and linked it. I only read the definition.
reply