I do this in OMP, a fork of Pi. It lets you set different models for different tasks. So with an API that has many different companies models I can set the Plan model to the best one, right now I am using GLM 5.2 for that, it plans really well. I have Vision set to Kimi 2.7 Code (cheaper and vision is just fine). Minimax M3 is set to the Advisor role (double checks work). Deepseek v4 Flash is set for the Task role. And MiMo 2.5 pro is set as default.
With this setup GLM handles planning and managing my AGENTS.md, and orchestrating subagents for tasks from the plan/todo GLM created. The tasks themselves are handed off to Deepseek v4 flash to implement with strong instructions and examples for each agent. Minimax M3 reviews the output as the Advisor and recommends changes, catches bugs, and whatnot, subagents can be re-run with that information.
Overall I am saving a lot using some of these smaller models. But with this setup I am getting great results.
Having watched Qwen kill its own llama-server instance to free up a port, I think this is a bold presumption and you should test it at your earliest convenience.
Reflecting on this for some reason reminds me of this passage from Kurt Vonnegut's "Sirens of Titans". I hope we use these tools to unlock something within ourselves rather than mindlessly expanding outwards.
"Mankind, ignorant of the truths that lie within every human being, looked outward–pushed ever outward. What mankind hoped to learn in its outward push was who was actually in charge of all creation, and what all creation was all about.
Mankind flung its advance agents ever outward, ever outward. Eventually it flung them out into space, into the colorless, tasteless, weightless sea of outwardness without end.
It flung them like stones.
These unhappy agents found what had already been found in abundance on Earth—a nightmare of meaninglessness without end. The bounties of space, of infinite outwardness, were three: empty heroics, low comedy, and pointless death.
Outwardness lost, at last, its imagined attractions.
I operate an analytics site (pretty big one B2B where client's backend feeds data into our system), and we see tons of traffic originating from northwestern China (Xinjiang) from Shenzhen Tencent Computer Systems Company Limited.
There are also half a dozen other companies from China continuously hammering our clients’ websites.
I was wondering, what's in that cold dessert? Low and behold satellite imaging shows massive datacenter build outs, very cheap solar energy.
Few months ago something happened and the Geo location on data on those IP now shows "Shanghai" or "Shenzhen". A way to cover tracks? But mapping latency still points to fact that nodes behind these IPs are still operating around Xinjaing region
credit:
'You Can't Cheat Time: Finding foes and yourself with latency trilateration' https://youtu.be/_iAffzWxexA
HN user: lopoc
Shenzhen vs Xinxiang is hard to do using this technique but Shanghai vs Xinxiang does show difference.
Assuming that China only distills is a huge mistake.
It’s no longer some backward place that does low value copying. Look at companies like ByteDance and Xiaomi.
Chinese companies aren’t just distilling, they’re acquiring data in the same way American companies did by paying people and crawling the internet.
The way I understand it, China has a few large companies that crawl the web at a rapid rate and build corpora. The government essentially wants select few companies to do this and then make the data available to other strategic companies operating within China.
Then there are data aggregators that buy data from apps, websites, and services, as well as systems like OpenRouter or Cursor, where companies can learn from the “traces” of coding agents, chats, and so on.
This massively reduces costs, as smaller companies like DeepSeek don’t have to do their own crawling or acquire data from 100s of websites and coding agents etc....
There are also companies in China that buy American LLM APIs and proxy them to companies within China. So, there could be 10,000+ companies using American AI products, while China logs all of this, understands how they’re being used, and trains on their traces.
People keep asking this but I don't see how this is even a consideration - it isn't going to have AI voice if AI didn't write it. To some extent humans have picked up some of these tells but part of the writing process involves reading your own work, noticing things that are awkward, and rewriting them. If you aren't investing in the writing process even that much, I don't really want to read it anyway.
My entire position: I'm not interested in reading text that sounds like it was written by AI.
Step back and think about it another way - ask which scenario is more likely:
Some random person discovered a 60% across the board gain in all LLMs, using an extremely simple trick that none of the labs noticed in all these years. That trick being to rasterize 8bit characters into 8x8 pixels in a big image. 60% in a market worth trillions of dollars.
or
Anthropic's marketing team arbitrarily prices tokens to drive growth, according to vibes and feelings, and didn't think they needed to price images on par with text in their rush to burn cash & drive growth. Some folks take advantage of the trick during the first few days of the model's availability before Anthopic corrects their pricing, to align more proportionally with actual compute costs.
Curious to hear if anyone has tried running the 2-bit or 3-bit quantization of this. With a bit of investment I may just be able to swing it locally. I already have 96GB VRAM, so with 192GB RAM, which seems to be the most one can find these days with a 4-slot motherboard, I may be in with a shot. Yes, it'd be slow, but I could give it overnight jobs. But I don't know if running at such a low quantization would make it hallucinate with only a small context.
Qwen and Gemma are great, but they need babysitting every 30 mins, which is quite a cognitive load.
For work, I mostly use Codex and some Claude. For personal use, I’ve started using Chinese models directly through their respective providers, mostly for automation tasks and experiments so far, either via the API directly or through the Pi harness.
I do not trust any of them. Everything runs inside virtual machines, not just the sandboxes provided by the harnesses. I also do not run Claude or Codex directly on the host machine. Not just because of supply chain fears, but also because of how incredibly user hostile the VC funded companies are when it comes to installing random stuff on your machine.
> During the 1700s, smallpox raged through the American colonies and the Continental Army. Smallpox impacted the Continental Army severely during the Revolutionary War, so much so that George Washington mandated inoculation for all Continental soldiers in 1777. Just fifty-six years earlier, in 1721, Bostonian doctors and clergy introduced the procedure to the American colonies. Without the vision and determination of these early Bostonians in normalizing inoculation, Washington may not have made the decision to mandate inoculation for the Continental Army. Though it was a controversial action, many historians credit the medical mandate with the colonists’ victory in the Revolutionary War and the creation of the United States of America.
> You are hereby required immediately to send me an exact return of your regiment, and to send all your recruits, who have had the small pox to join the Army. Those, who have not, are to be sent to Philadelphia, and put under the direction of the commanding officer there, who will have them inoculated.
I'm happy to give my identity to Anthropic and crush my competition with irrational fear about privacy and personal data. This is a serious competitive advantage and a moat.
I have a home server that runs Qwen3.6-35B-A3B through llama.cpp with Open WebUI for the user facing interface.
My teen isn't super interested in AI, but whenever they do feel curious they have their own account they can use on our home network. As far as chatting goes local models are more than capable for handling standard chat questions, doing research, helping troubleshoot problems etc. In fact it was an agent powered by the same model that setup the open webui server and took care of all the account management features through my phone (using Hermes agent).
If you're building AI powered features and using sophisticated agent setups for coding for work, then it make sense to use SoTA from these providers. But I've been using local models increasingly for personal use and am starting to find them preferable (I run an uncensored, ephemeral model for my own use and it's an entirely different experience than anything you can pay for).
Still haven't cancelled my personal Anthropic subscription, but considering it soon.
Unfortunately, in the USA, we are stuck with two parties. Measured along axes of competence and capability of taking action: The (D) party is competent+incapable, and the (R) party is incompetent+capable. So every four years we get to choose between governance that knows what it is doing but too paralyzed to act, and governance that doesn't know what it is doing but acts with impunity.
I get that many people are tired of too many AI posts and there's an influx of negative comments but I think this is genuinely amazing. Think about watching Star Trek in the 60s and seeing people talk to computers and them understanding and being able to communicate back. We're literally living in the future!
It’s all just vibe coders high fiving each other and saying LGTM.
Here’s a secret: I haven’t really bothered reviewing any PRs since September of last year or so. I just click approve and don’t even read it. It has been way less stressful for me.
Do bugs get through? Sure. But someone will just come in and vibe it out anyway. And I have yet to see a vibe coder or anyone who approved their flawed work get any repercussions. So yea, it doesn’t matter.
No one’s going to come after you asking “why did you approve this shitty PR?” And if they do you can forward it to the author and ask why they wrote a shitty PR. But that just doesn’t happen. It doesn’t matter.
Correct. I use AI a ton and I'm having more fun every day than I ever did before thanks to it (on average, highs are higher, lows are lower). Your characterization is all very accurate. Thank you.
Im gay and because of that was disowned. My partner has a brother “K” and K has three children. Watching K show up in basic ways for his kids, like remembering what songs they like and teaching them sports is the fastest way to make me ugly cry.
Thanks to anyone reading this if you’re trying to be a good dad. You’re making the world a better place in ways you don’t even see
Being able to write well is arguably one of the most important things any person can ever learn, as you can't write well without thinking well.
So, I immediately write-off anyone who produces their writing with LLMs, as it is quite apparent that they are not a serious, thoughtful person and are instead just cosplaying as one via LLM. The fact that they can't see this (or, worse, think the slop actually makes them look good) only further proves the point. I'd much rather read flawed but genuine writing over vacuous AI slop, and surely most people are the same.
The other day someone posted in reddit about a repo that they seem to have put a lot of thought into. But the post and all comments were AI slop. I told them that while the project looked promising, I couldn't take them seriously on account of the slop writing.
They replied thanking me. Then the next day followed up to say that I changed their life - no more Ai-generated writing when interacting with other humans. I hope they were serious about it.
Are you using multiple agents on the same project? If you are, use git worktrees.
I personally prefer 3-4 concurrent terminal sessions, working on different projects. Claude code usually runs around everywhere with agents, and sometimes takes 20 minutes to get shit done. During that time I'm shipping open source work, etc.
I use my own tools to help: Sugar and RemembrallMCP.
Sugar is used for memory that's stored outside sessions. So I say store this in sugar memory. If I open a new claude session later, and ask it to lookup sugar memory, it's there.
RemembrallMCP is a AST and code graph that helps the agents understand change impact of your codebase. Instead of insane greps veritical and horizontal, it uses this to quickly understand impact with very high accuracy, time reduction and low token limits.
German General Kurt von Hammerstein-Equord (a high-ranking army officer in the Reichswehr/Wehrmacht era):
“I divide my officers into four groups. There are clever, diligent, stupid, and lazy officers. Usually two characteristics are combined.
Some are clever and diligent — their place is the General Staff.
The next lot are stupid and lazy — they make up 90% of every army and are suited to routine duties.
Anyone who is both clever and lazy is qualified for the highest leadership posts, because he possesses the intellectual clarity and the composure necessary for difficult decisions.
One must beware of anyone who is both stupid and diligent — he must not be entrusted with any responsibility because he will always cause only mischief.”
Sbf built a pyramid scheme that collapsed. Sam altman has built a pyramid scheme that will soon collapse. The collapse is the point. This way insurance companies, advertisers, etc can buy the company and its assets, the customer data, that gives them a level of access to people laws and anti-trusts would otherwise prevent.
“When I was 15, I spent a month working on an archeological dig. I was talking to one of the archeologists one day during our lunch break and he asked those kinds of “getting to know you” questions you ask young people: Do you play sports? What’s your favorite subject? And I told him, no I don’t play any sports. I do theater, I’m in choir, I play the violin and piano, I used to take art classes.
And he went WOW. That’s amazing! And I said, “Oh no, but I’m not any good at ANY of them.”
And he said something then that I will never forget and which absolutely blew my mind because no one had ever said anything like it to me before: “I don’t think being good at things is the point of doing them. I think you’ve got all these wonderful experiences with different skills, and that all teaches you things and makes you an interesting person, no matter how well you do them.”
And that honestly changed my life. Because I went from a failure, someone who hadn’t been talented enough at anything to excel, to someone who did things because I enjoyed them. I had been raised in such an achievement-oriented environment, so inundated with the myth of Talent, that I thought it was only worth doing things if you could “Win” at them.”
It's a fools errand to ever believe in job security. Even if you're absolutely right on your importance, management can always remain stupid longer than you can remain employed.
To stop this, I today put most of my Amazon Redshift research web-site behind a basic auth username/password wall.
It's all remains free, but you need to email me for a username and password.
If I put in time and effort to make content and OpenAI et al copy it and sell it through their LLM such that no one comes to me any more, then plainly it makes no sense for me to create that content; and then it would not exist for OpenAI to take, or for anyone else. We all lose.
With this setup GLM handles planning and managing my AGENTS.md, and orchestrating subagents for tasks from the plan/todo GLM created. The tasks themselves are handed off to Deepseek v4 flash to implement with strong instructions and examples for each agent. Minimax M3 reviews the output as the Advisor and recommends changes, catches bugs, and whatnot, subagents can be re-run with that information.
Overall I am saving a lot using some of these smaller models. But with this setup I am getting great results.