I run lmstudio personally with a range of harnesses (open and closed) and can't say there is that much of a leap to getting everything talking https://lmstudio.ai/docs/integrations
To me this looks like another case of bundling things that shouldn’t be bundled (the harness with the UI) making both worse off because you can’t individually focus on each component. They’ve done this before, bundling a decent UI with decent inference for IMO no good reason, combining the downsides of each instead of letting people mix+match.
and if someone can't figure out how to write down an address it's very likely they also can't figure out how to make local models not suck for coding, and would likely switch back to codex/cc after 15 minutes anyways.
I agree. Incidentally, this is exactly what ollama are doing too.
Since most people are unaware of this fact.
Happy to clarify which is who and who is which.
Less unanimous and debatable, but many would say they more often do not align than the opposite.
Historically it was not. They are two mostly unrelated things. Creating more with less labor can mean the richest getting richer while middle class becomes poor and poor become hopelessly poor.
> Creating more with less labor means a society of relative abundance, with more time for education, creativity, leisure and personal growth.
It never meant that, this is fairly tale. We got 40 hours workweek rather then 80 hours workweek because of political movements and fights, not because of technology. Labor saving device, on itself, leads to two outcomes: you work as much as before, but produce more (cue all the burned out overworked ai coders) or you get unemployed desperately looking for new work.
Production and consumption are directly linked: societies can consume only what they produce. When individuals become more productive, society has more to consume. And indeed there is more. So much more. What you call a "fairy tale" is, thankfully, the world we live in.
When considering prosperity, it's important to emphasise what's real. So what if Warren Buffett owns a massive number of shares in notionally highly valued corporations? There are plenty of carrots available in the supermarket, and at an extraordinarily low prices by historical standards. Warren Buffett is not personally consuming some grotesquely disproportionate share of the carrots ... or pretty much any commodity which constitutes real material prosperity.
Nor were they result of a new abundance.
> When considering prosperity, it's important to emphasise what's real.
Yes. And what did actually happened in the past.
> So what if Warren Buffett owns a massive number of shares in notionally highly valued corporations?
Meanwhile, Warren Buffett : "There’s class warfare, all right, but it’s my class, the rich class, that’s making war, and we’re winning."
The accumulation of power and money by smaller group of people means loss of power by the rest people. Those with power then use it to their own benefit while rest of us dont have that option. This result in the small group having even more power, while needs and wants of the rest of us are ignored. In turn, it means worsening conditions for the people on the loosing side.
And you see this process happening as we speak here. You see it at political decisions being made by the administration, at climate change approach, at unaccountability of the Epstein group of friends, at wars for oil, at decay of the democracy.
Materially you see K shaped economy, in us being unable to buy RAM and electronics, loosing rights, loosing control of devices to the corporations, basic necessities getting more expensive and so on.
The RAM example is particularly absurd. Two decades ago, ordinary consumers were not complaining about the price of 64 or 128 GB of RAM because owning that much was barely imaginable. You are complaining about access to a product that exists only because capitalism drove extraordinary technological progress. Obsessing over RAM shows a lack of historical perspective and a badly distorted idea of what constitutes a good life.
Yes, concentrated wealth means Warren Buffett has a disproportionately large share of influence. But that does not mean Warren Buffett also consumes a disproportionately large share of the resources that represent human flourishing. Ownership, political power, and physical consumption are different things. Corporate shares are not sandwiches ripped from the mouths of children.
And blaming every disliked event, from wars to Epstein to electronics prices, on conspiracies involving the ultra-wealthy is exactly the kind of unhinged, zero-sum populism that corrodes democracy. Instead of evidence and analysis, we're given a third rate morality play about a wicked elite stealing everything from "the people". Extreme polarisation and populist thinking are corrosive forces in modern society.
It wasn't the machines that caused the social progress. It was the organised labour movements (and the variously destructive or violent sabotage movements) pushing back against the machines that made the social progress.
The tractor and the loom made life noticeably worse for the people they replaced until they organised against them, and even then it took decades for them to get back the limited kind of quality of life (and health) an agricultural worker might have had.
I do find it interesting how many AI people make tractor and loom analogies without the slightest grasp of the true history of the industrial revolution.
Any social progress that comes from AI will not be something AI people can claim as their success, because it will have happened at their expense. There will be pushback; it may be violent.
As for the rest, it's a remarkably selective account of the industrial revolution. Your argument relies upon the unspoken inference that pre-industrial agricultural life was somehow a better life that was diminished by progress.
The tractor did not merely replace farm workers. It allowed a tiny fraction of the population to produce vastly more food while releasing millions of people from exhausting agricultural labour. Perhaps you would prefer a life of hard physical labour, low output, poor medicine, food insecurity, and little protection against injury, disease, or harvest failure. Perhaps you don't like that you toil away in an air-conditioned office writing video game code, for a salary that you wish was higher, because you dread the cost of your next smartphone purchase.
Technological change created enormous productivity gains. Yes, these often unfairly disadvantage workers trained in the outgoing industries. But I don't hear you arguing for a return to hand-weaving, because the societal benefit of progress is inarguable.
I am not at all suggesting it is necessary; you are projecting that onto me. I am merely observing that it might happen. Already is happening.
> Your argument relies upon the unspoken inference that pre-industrial agricultural life was somehow a better life that was diminished by progress.
Well for several generations life expectancy actually fell, working hours increased, general health worsened, child and infant mortality worsened. It took until the 1900s for those measures to improve. Child labour became an outsized evil until work and education reforms happened really only 140 years ago.
Do you not understand that industrial work was not less exhausting until workers rights movements did something about it? Not at all less dangerous? Not at all less dirty, less violent, better paid, more free, more secure?
For the first century or so, the industrial revolution (and the second agricultural revolution) made the average worker's life worse — less stable, less constant, fewer rights, a destruction of traditions and social cohesion.
The social progress you argue that technology drove, it drove as largely a negative influence. Did we get modern medicine after two hundred years or so? Of course. Was there industrial progress? Of course. But social progress happened as a result of the labour organisation of people to stop their lives being made radically worse by a class of new extremely rich capitalists. Productivity gains are not social progress, nor are they a proxy for it.
> Perhaps you don't like that you toil away in an air-conditioned office writing video game code, for a salary that you wish was higher, because you dread the cost of your next smartphone purchase.
This is childish straw man projection of which you should be a bit ashamed, IMO. That is the kindest thing I can say of it.
Claiming that life expectancy "fell for several generations" is a poor reading of the data. Most generously there was a period of stagnation predominantly caused by a combination of scarlet fever and rapid urbanisation.
More generally you're arguing against positions I haven't espoused and which I don't think even represent areas of disagreement. Productivity gains and social organisation are not opposing forces. The surplus from higher output made shorter hours, better pay, public health, and medicine sustainable, making it possible for labour movements to secure larger shares and better terms of that surplus. We live in an amazing world now, if you take a moment to stop and think about what actually matters.
It’s an important criteria to have in mind when you select an application.
and using a closed-source, VC-backed app that might change anything in the next update might not be best for privacy
Is that a problem for you comrad?
I don't think we need closed-source developer tools, especially ones where they might restrict access if they decide to start charging for them later.
I would obviously prefer an open source, open weights stack.
But I guess a paradox is that as long as there are open source options I could use, a solid agentic environment that I can use with my own open weights is something I might pay for, in a similar sort of way to paying for a Mac when I could use only Linux.
If someone wanted to make their entire income from, say, making the BBEdit of LLM harnesses, that would be a viable strategy. Sooner or later people need to make an income somewhere. My own feeling is that Apple should acquire LM Studio, but if they said "this is $X per year" I might consider it, given the attention to detail.
The M series chips provide large amounts of GPU RAM so I think people can see a world where Apple sells local first, privacy focused, inference based on hardware sales.
LM Studio's strategy seems to fit that very well.
So they put revenue above true privacy, as companies do. But they are the loudest to spit the bs while selling black boxes.
I'm not pitching Apple as the best or only solution, I'm saying that selling local inference seems to fit with their current marketing goals.
Selling high end, high margin computers that have lots available GPU RAM with at least lip service to local inference for privacy fits with what they already do.
I don't see anything to stop say Framework making the same pitch but with Linux but since Framework don't design their own chips they don't have quite so many levers to pull as Apple do.
I think it ought to be possible to understand my comment without assuming it is Apple fandom; it’s not.
(see e.g. https://news.ycombinator.com/item?id=48805487 for how I feel about brand loyalty)
I am just one of those developers who use Linux daily as a remote/VM work environment, from a MacBook, from personal choice. I have for decades. There is a kubuntu laptop on my desk too (a Framework) which is fine; I have more or less always had GUI linux around too. But my own LLM dabbling is happening on an M1 Max, like so many others, because I am not a gamer or home build PC guy (this century at least) so I don’t have a PC with a GPU. (Yet. I am LLM-cynical but I can see running a dedicated machine for this in principle)
The suggestion about Apple (who I don’t believe are yet “out of the picture” in any meaningful sense) acquiring LM Studio was an idle reflection on them being the most Apple-ish of the options, that’s all. They need something; LM Studio is closest to what they need.
i believe that for most people on the street, for most tasks, a Chat GPT 3.5 era LLM is sufficient enough. sprinkle in tool calling and other things, and that becomes enough. if you can prioritize that level of a model on-device (baking it in etc), then you can bifurcate AI users between those unwilling to pay and those who are willing to pay A LOT for frontier model performance.
Nothing has changed at the level of computation either, hardware speaks the same vocabulary and of electrical signals as ever.
Could you describe the change you are observing?
Tools like Opencode demonstrate that when you box them in tightly enough they can actually be pretty competent.
That way everyone has access, even with older devices, and it's a subscription! Then Apple can tie their APIs into the ecosystem you love at a flat cost you can afford. No need to support local model integration in the first place, problem solved.
Please don't slur us older GenX as boomers.
It already is.
This is why I think it will be the platform owners who end up winning - if iPhone / Mac / iPad ship with a good enough local model that can directly interact with the OS and apps and the web people might actually start talking to their computers.
Most people can barely type or operate a computer, so I think it will be a boon to many folks to have a voice/agent interface to their computing. I agree that a local model that's "good enough" will suffice for most general tasks. There's no need to call Mythos to "add cheese to the shopping list" or "summarize today's emails". Harder tasks will call out to stronger models, but many people just won't need that power.
I suspect all the platforms are keenly aware of this, and Apple will drop Siri v2 or whatever when they're satisfied with their UX and more confident the agent won't go off the rails.
That said, and this is off topic: Siri on the newest iOS beta is surprisingly good now. I asked it what model it was using yesterday and it said Gemini for difficult problems, then secure Apple model in cloud, and local Apple model.
https://runaihome.com/blog/wwdc-2026-apple-ai-home-lab-verdi...
(I’m the founder of LM Studio)
> use the largest frontier open source models through LM Studio Secure Cloud
Edit: If you keep checking back, I'll have a video showing off that application. It's a bit beefy to just get up and running.
For some of the latest big models (from Minimax, GLM, Kimi, etc.) even after months they still don't provide official downloadable versions, just their cloud one.
There is no way I’m running a python or JS agent which is probably vibe coded, it’s just too much risk from a security and supply chain standpoint.
I use ghostty despite knowing mitchellh used AI to build it, because I know he's not vibe coding it. I don't use any Anthropic software for the exact opposite reason.
Try it out for coding (in a "Code" project) and document creation / manipulation (in a "Work" project). In Work projects we have automatic checkpointing for every change the agent makes. Would love to hear your feedback.
They do update that but with lag
This is one of the better agent harnesses I've seen for inspecting reasoning chains, which is super useful for me. Sometimes reading the reasoning is better for my needs than reading the response. I appreciate how transparent that is in contrast with Claude Code/Codex/etc.
One question — I saw you mentioned negotiating ZDR with your "providers" — are you hosting the models yourselves, or is another entity hosting them? If another entity, which group?
I had a quick browse through your site, but I could not see where you are based, but I get whiffs of more American 'trust me, bro' exports which I've been eye rolling at for two decades now.
First impression: it works great. I use Codex as my main agent, and the UI looks similar enough that it's familiar and simple to get started. I just pointed it to my existing LM Studio models library, ran Qwen3.6 35B, and the results are exactly what I would hope for.
I did notice some rough edges that might be worth improving, however:
- Current working directory is not the clearest on the main page of the app. It shows the project name, but is missing the prominent working directory label like Codex has. - The model seems to load when you hit Enter, but it shows "Working" instead of of "Loading model". - There doesn't seem to be a way to pre load the model, it seems like you have to send it something to load the model. - I don't see a way to easily unload the model like the eject button in LM Studio without quitting the app - I pointed it to a directory called "GitHub & Projects" and it somehow ended up making a new folder called "GitHub & Projects". Yes, I know the name is weird but it shouldn't have done that.
* locked to a single dir, so no system wide access.
* no local web search, can be fixed ddg or local mcp.
* no ssh, I want to have it ssh into my server and do the work.
* doesnt show the model being loaded, needs a bar/% counter.
* Can you drag/drop documents in the work dirs, or only + add them?
I love lm-studio, so cant wait to see how this goes. For local I normally use opencode + lmstudio.
Am I blind?
The word "Bionic" is never mentioned on the homepage
Any other ideas?
I wish LM Studio played better with AMD hardware. It would be really great to have an off-the-shelf solution that "just works" on Radeon.