The deterministic check could help you filter out the chaff - prevent the human element from getting DOSed.
3,881 karma · joined February 15, 2019
Also, this has another angle, let's say I've created some tools I've invested quite a bit of time in but what is the point of publishing it, if anyone that needs it can simply make it for themselves (spending a week or two of time and some tokens)?
Not only it has became essentially impossible to find if a given thing is quality (for example I recently looked for a pi plugin that does tool use classification for AI that I could extend with my ideas. I found one, then I discovered the author copied leaked cloude code source and "released it under MIT license" - which made my work on top of it basically useless and I had to throw it all out and start from scratch, as I wanted to publish the result- the final thing is called adaptive-acl).
And second, is there any point in publishing code? In the "old days" when software was worth something the motivation to publish open source, in addition to feel good factor, was that people would send you patches with functionality you may find useful.
Both that and the feel good time has ended. I still publish some of my stuff that I think may be useful to someone at some point, but I'm not sure there is any actual point in this.
In a way AI has made each of us much more self sufficient. A computer with a good local model and half a dozen GPUs, few TB of software and one can have all of their computing needs fulfilled forever. No Internet or other people required. There is certain power in this. My young pre-internet self would have felt like living in a star trek universe with this :-)
But something is definitely being lost in the process as well.
Seriously? Discovering proxies exist and actually finding the people that use them are two different things.
Trying to conflate those two things in an article meant for a technical audience is quite funny.
In fact I'm pretty sure it's happening. Why? Because of constantly growing prices of those things.
I was genuinely excited about this device until I heard the specs. Still under 300GB/s memory bandwidth when a gpu like rtx3090 from 6 years ago had 900+
This is really bread crumbing local inference users.
We should be measuring our memory bandwidth in TB by now. The only company that seems to be serious about this is Apple. (the m5 ultra studio etc).
You're literally replacing a well designed resilient system with a single point of failure run by a different company...
But this would require an admission of limitations of the technology.
Do people sue Google if search links to a page that has misleading info that results in someone taking a wrong drug? Nope.
Do we sue the browser maker when the info on a given Web page is malicious? Also no.
So, I don't understand why we treat chatbot providers any different. Everyone should be expected not to believe a word they say without corroboration.
Good luck starting a fab in Europe... Maybe "hobby sized" attached to some university. But certainly not anything that can make parts in volume.
It used to be the case China competed mainly on price thanks to low labour costs. These days they still compete on price but because of supply chain integration.
If like me you dabble in hobby grade electronics you know how parts and services availability looks like here in Europe. We have only one component distributor now that UK left the EU. And even them (Mouser) if you try to order any advanced components that run at high frequencies etc tend to ship stuff from US warehouses.
Dont get me wrong. I'm still amazed they can get an order to me from US to Poland for $15 and 3-4 days time. But if I lived in China I could walk to one of bajillion physical suppliers and have every component made available same day. If I don't want to overspend on my prototype I could buy used parts without having to wait 3 weeks for delivery.
Then we have PCB manufacturing and assembly. It makes zero financial sense to use these services domestically. I'd love to support local EU industry, but I'm not spending 10x the cost for the privilege.
I find well prompted AI makes nice looking websites. Are they better than some designer could do?
Probably not, are they better than a guy that hasn't done a minute of fronted work in 20+ years in the industry? Probably.
You can give it instructions and 5k tokens later it will do exact opposite you told it to.
It uses all if my 6 rtx3090s but runs 7 sessions at combined 200tok/s and prefill of 7k on a cheap pc (no threadripper, all gpus via usb4).
DeepSeek v4.1 Flash 1017 is better but it is also 3x the size... And I find running quant below 3.8bits unusable.
But flash next is good enough (at 4.5bits).
In fact I fail to see any other end game where they make profit on the huge liabilities that now have.
Those products are a sick joke played on us. About the only thing they have that is better than tjose ancient gpus is energy efficiency.
And what good are the weights when all the rtx3090s fail (they were pretty crappy to begin with) and every ddr3 Xeon system fails too?
I'm fully prepared to be running on my hardware that is 6 and 8 years old for next decade. Last 3 months saw rtx3090 going up by 20% each month.
I've been running something similar for few weeks, but with Qwen3.8-Flash-Next. I started with opus 4.8 limited to 200k context as the supervisor and 5 Qwen3.8-Flash-Next agents(262k context each, autocompacted). A very simple way of calling agents via a command line using pi.
Telling agents to maintain task status in files.
This has worked very well. I tried Qwen3.8-Flash-Next as the supervisor too and as long as the plan is written by a frontier model all the implementation can be done by cheap/local models.
When I have lots of code to write I use DeepSeek v4.1 Flash 1017, because my entire local Qwen3.8-Flash-Next throughput is 200tok/s across 7 sessions, while a single DeepSeek session runs at hundreds. 7 DeepSeek agents and one Qwen supervisor can complete substantial tasks fable needed 12h in 2h.
I haven't tried Qwen3.8-27b for coding. I've been using it mainly for interactive chat with Web tools via librechat. My benchmarks show it as weaker than Qwen3.8-Flash-Next by few points.
For example the laws treat the output power of the inverter, the kWh capacity of the battery and peak kW capability of solar panels as a single number and they simply call it "capacity".
The law states your "capacity" cannot exceed the power contracted with the grid provider.
So if your home has 20kW of power contracted from the grid you're not allowed to install a battery larger than 20kWh,or more than 20kWp solar, or your inverters can't have more than 20kW. This results in an extremely lopsided systems.
In our climate a good system for a house like mine that uses about 20kWh a day, and has 24kW of grid power contains 14kWp of solar, 65kWh of battery and 20kW of inverter power.
Other examples of stupidity. Any inverter above 3kW has to be 3 phase or it needs to be a part of a 3 phase system. Why? What is the point of that? Most 3 phase inverters allow up to 50% phase imbalance. So if you have 12kW inverter you can't use more than 8kW on one phase and most are worse than that.
Furthermore and this is the pinnacle of idiocy. Any system with over 10kW capacity (either solar, battery or inverter) has to be managed by the grid. You pay for them to manage your own system and the electricity company will decide when you buy electricity from them and when you use yours... Guess whose financial benefit they will work towards.
So what is the solution? Put our houses off grid and that is what I did.
I have a solar setup that runs my house off grid for up to 8-9 months a year, for the remainder I switch a transfer switch and I run on the grid.
How is this different from using your €90 Anthropic subscription to the fullest and then discovering same usage via api would cost many thousands a month.
No doubt they are loosing great deal of money on it, but if they provide this service why should an average person not use it?
Even things like decommissioning an old model can be extremely disruptive if all your workloads are tweaked for it. Like when Opus 4.8 dissappear from copilot in one company I work for. Suddenly our PR reviews started missing issues they always caught before.
Of course I never expected anything different from the Chinese Communist Party.
Qwen3.8-Flash-Next consistently benchmarks better than both GLM5.3-Flash and the previous DeepSeek v4.1 Flash 0731. The DS 0731 was so atrocious it was getting beaten by small Nemotron in my coding tests (subset of tasks from open source coding benchmarks).
Now, new DeepSeek V4.1 Flash 1017 really gets the numbers they promised for the first one, but it's too big for me to run the 4 bit quant locally at interactive speeds.
So I kept Qwen3.8-Flash-Next as my go to.
Until it got lobotomised. Then it lost most of it's capability.
Similar situation happened to Anthropic's Fable. It was an amazing model until Opus 5.5 came out.
So now, I'm not even looking at a "new" model until it is a month or two old. By then we will know if they will destroy ability to make it cheaper to host.
Why do you need to protest at the slightest sign of the government becoming authoritarian when you have "the ultimate argument of force" on your side?
But it's an illusion. Guns are a nice hobby I enjoy as well. They are great for self defence, but good luck to you if you think you'll use it "against government tyranny"... You'll be dead and no one will care.
2nd amendment makes an ultimate duty of a citizen in a democratic country - to not allow it to turn into a tyranny, into "you don't need to worry about politics, worse case scenario you have your AR15 and a box of ammo".
While in a country without such false ideas people get out and protest in mass numbers.
I don't think I've ever heard of a proper protest happening in USA. The kind of protest that involves 10% of adults at least and happens in every city and town. (Also I'm not talking about rioting like in France)
Burning down a school to protest against class overcrowding is like something straight from idiocracy.
Peaceful protests work on governments, even those hell bent on authoritarianism, but they have to be massive.
Let's say you have a million or two of people show up in Washington. Will it make any difference? Nope. Why? Because it's under half a percent of US population.
Let's now say 10% of people in every city and town go to their town centre and protest. Does this make a difference? Yes it does, because it shows real power to everyone.
But in US human life is not worth much.