> I can’t get Claude to do anything even a fraction of this complexity.
I remember writing a postgresql compatible DB with Opus 4.5, that used S3 as storage and local caching to make it speedy.
Ironically, Opus 4.5 is by todays standards is antique. If you have some knowledge about Databases, it goes a long way.
But you need to do it step by step. Getting the core to work, getting a Pratt parsing going. The whole pgwire protocol ... the data format ...
Step by step ...
With todays Models, your can probably get away with using /goal and telling it to make a postgresql compatible database, while having it run a few days.
Now, making a fun database test project and having it production ready! Big difference!
> If you're acting in good faith and your accountant does something crazy or evil, your liability is limited to some extent.
From my understanding, you are the person signing off on the paperwork that is submitted to the IRS.
There is this cache 22 with taxes.
You are responsible, but you outsource it to a accountant. Because you are not knowledgeable about the taxes. But you are expected to be knowledgeable to understand the tax documents that you submit to the IRS. That is why the accountant always ask you to review the documents and sign them like 20 times.
The same applies when you run a company, depending on the country, you need to prove yourself knowledgeable in accounting, before you are allowed to run a company. Normally that is included in a university degree, but if you have a middle school diploma, you need to do a official examen to get that degree.
Whatever you submit for your company, you are again responsible. Even if you hired a accountant.
So while technically, if a accountant makes gross mistakes, the bill will always fall in your lap, because you are expected to understand the reports you submit to the IRS. And catch any errors before doing so.
With the IRS, the burden of proving your innocents is often put you. Its because the good faith argument can be misused easily. That is why the buck stops at you.
So using a LLM or a accountant, really does not matter. Sure, a accountant can go to jail if there has been major issues (its not going to be with one client issue).
But you can lose your house / company, have your life ruined by whatever you submitted.
O yea, that is driver lane assist ... A Toyota rental had the same issue. In a specific steep exit corner (that goes up facing the sun), how many ** times the lane assist tries to force the car to go straight (as in, off the hill! ). The first few times when it happens, scares the ** out of me.
Another fun one is going down a hill in a Rental Opel, roundabout with some cars, no problem. Slowing down naturally, while i see the cars accelerate to enter the roundabout. No need to break as by the time i get close, the cars will have started to accelerate. So my speed will have matched the last vehicles speed by the time i am close. Suddenly, emergency break slam on !!! Because "the car was going to hit the cars in front". Like, wtf!! That created a extreme dangerous situation if there was a car behind.
I really see no benefits for a lot of those new safety features. The old ones like traction controle etc, great, keep them. But all this external monitoring, internal monitoring ... If your a safe driver, those features can make it more dangerous.
> If you measure "function" by the relative economic and military power of the country, then the EU has overwhelmingly degraded the function of its initial members compared to when they joined.
Very sure that when the EU was still in its infancy, we had only "west Europe" in arms, vs a USSR (aka all the eastern states and Russia). Now all those states are part of NATO and the EU.
Instead of the border to the closest hostile nation (Russia) being barely 100km from here, its now over 1200km to the first contact point.
That same Russia can barely deal with a Ukraine, that has some spare change backing from the EU. How is again at a war economy? Ukraine, sure, Russia, sure, EU ... nowp.
We now have Northern members that used to be neutral or not part of NATO, that are now part of it.
I feel like people love to misrepresent a lot of history. We have never been in a better position as a EU, vs what we used to be 40, 80, 100 years ago.
Yea, we have a lot of buildup to do again, but lets be honest, i rather see buildup now with modern kit for the modern battles, then relying on outdated 1990's doctrine and weapons. And even that is still a slow process with transitioning to the new reality of drones, drones and drones. Do not forget that 90% of the kills are now by drones.
People love to parrot those US talking point that often have no sense of history and our current EU reality in regards to security. While i admit, that we are still too reliant on US kit, even that is slowly changing. The EU moves slowly but it moves. Better then being some nations that are stuck in Imperialistic ways of thinking, like Russia.
> I'd be very surprised if it wasn't more than 50% cheaper for nearly all workflows, for a very similar level of quality.
If your using pure API ... providers like neuralwatt cut that cost down even more by using energy as the actual cost. So GLM 5.2 is more expensive then GLM 5.1 on their service (those thinking tokens), compared to API costs, its dirt cheap. And way more tokens then the zai subscription delivers.
We are seeing a move towards more realistic pricing on actual consumption based usage. Be it DeepSeek, Xiaomi (MiMo), or zai's GLM via neuralwatt.
The main issue facing subscriptions a-la-carte usage, is that a lot of the heavy hitters really drain the resources. And that as a business model can not survive without ...
a) increasing the prices.
b) everything goes to actual token/energy usage based billing but with more realistic pricing, and not the bloated API prices that are focused on companies.
We shall see what the future holds but things will change.
I like quiet keyboards but finding non-laptop / non-membrane keyboards that do strip the leather from your wallet, is a almost impossible task. Let alone one that does not grow in noise level over time!
There is way too much focus on a entire enthusiastic click click keyboards, but everything that i keep finding, sounds like absolute horrible loud, in a quiet room, where your partner is sleeping in.
And when you find something, its often a import, not in stock, and you get presented with like a $300 bill and no guarantee about quietness after a long time using. Or worse, they changed the keys in between production runs, and its now more louder.
Why am i writing this? Don't know, maybe tired of often wasting hours seeing youtube videos and reviews, and posts about keyboards to just feeling burned out.
> AI are unteachable, if you have given them a good prompt and they do something wrong 90% of the time you are shit out of luck.
If the Model makes repeated mistakes on the same subject matter, you can update your agent.md file, or you can add skills to deal with specific prompts, or you provide a better default harness.
The whole idea of coding agents is their harness makes a big difference vs a pure raw model.
> However AI cannot meaningfully handle feedback and learn
How do you think models are created? They are trained on feedback and learn.
Its not cheap but you can post train models. This is how custom models are mode, that deal with specific tasks more efficiently and accurately.
Example ... Composer? Its base Kimi v2.5 model that has been post-trained 2 weeks, to create Composer 2.5, what is a much better coding model.
Its literally trained to make less mistakes by feeding it correct data. Hell, a lot of the models you are using, are often the same base model, where v2.0 was the initial released model but the model keeps training, so when they release v2.1, its still the same model, but with more training time on feedback provided to v2.0.
LLM Models are not a cake you cook one time and they are done, and you start from zero again. If you have the money, and a powerful server setup, you can take a model like GLM 5.2 and post-train it, to reduce specific errors. Sure, you need a ton of money because its a large model.
But people have been doing this with 5M, 100M, 1B, 5B models for a long time already. To the point that some of the small models can do specific tasks, almost or better then some of the huge more general trained models.
> they sell you a subscription for frontier models, then over time begin to nerf them rapidly and no one talks about it.
People talk about it all the time. Just check some of the dozens of forums where its non-stop complaining about nerfs, limit nerfs, performance issues etc...
Is hard to prove that any downgrade is a effect of being deliberately served a lower class model / lower quant, or whatever. Or the "optimizations" hurting the models performance.
The TOS allows for those service "optimizations", so legally, nobody has a foot to stand upon. Like when OpenAI or was it Anthropic played with the cache, this to free up more server resources, only to later discover that its gutted the long term context behavior, and heavily degraded the models as context grew.
If you want 100% guaranteed the same performance/behavior, you need to run a model yourself (be it rented GPUs online or your own local setup). But its going to cost you a lot more ...
If i was going to write software on this level, that will be used by governments. There is no way, its going to be a nice little program that only extract information.
Its going to have every trick in the book (and outside it), to stay hidden. And it will have payloads to alter its behavior, updates, etc...
Nobody is going to pay you big fat money envelops for software that anybody can write in a afternoon. You want it to be as capable as ever, and you do not want it found!
> As far as we know that directive has remained in force with each successive administration as well.
People can state a lot, as long as your not caught.
Nothing prevent you from having the UK spy on the Germans, and feeding that intel back. Or Israel, or ... Hey, the US did not spy on a EU ally. Well, not directly and it neatly bypassed any official statements.
They might have simply gone to one of those secret court hearings and have it bypassed with a gag order in place. Officially its not done, unofficially, its been approved.
The whole "as long as you do not tell me your doing it" approach, and the politicians involve maintain deniability (even if they had the wink).
And you do not need to specific target the head off state. Plenty of side routes to still get information on meetings, that involve those heads of states. Even if your not "directly" spying on them.
So no, its a naïve way of thinking. Maybe in 20 years from now we find out, that they did spy on EU leaders. Maybe directly, maybe indirectly ... even with that directive in place. I will be amazed if they did not. Its the US we are talking about.
Fairly sure that if anybody using a advanced piece of hacking software, they are also going to delete any messages that are related to detection of such hardware.
PC viruses used to do that stuff going back so many years ago. Suppressing any notification under Windows, by disabling the AV software, its notifications, windows notifications related to it.
So it will amaze me that this is not done by any modern espionage software. Especially as the notification methods are known. Given that his device is hacked, that means a lot of avenues are under control of the espionage software. Even mails etc ... So impersonating the end user, to confirm they read a warning, is extreme easy.
I find it rather odd that people are so fixated on the idea if Kouloglou read it or not.
I think it makes houses/apartment less likely to sell. When you see the idealized version, and then the reality, the impact is much bigger then just showing reality.
Unless people prove me wrong, and they really fall for that...
Its like we used to be flooded with fisheye lens pictures of homes, that made the rooms way bigger then reality. I noticed that this trend (on the immo that i follow for years) has heavily reduced. Because nothing beats a sale, as people seeing something looking spacious on pictures and then in person seen its way more small/cramped/compact.
I love that new trend of 3D home viewing... It saves you so much time, and saves time for the immo people, filters out a lot of people with less interest.
Am i missing something? Because your making my point. Its only worth it compared to Opus 4.8, if the tasks your running requires Opus 4.8 low (or non-existing lower).
For the rest the gap in pricing vs efficiency is so small, that there is no point in using Sonnet. I am looking at their own cost comparisons vs efficiency...
Anybody notice that they did not include Sonnet 5 Max in the "Agentic Search results", when comparing to Opus 4.8 ...
Based upon the "Agentic Computer usage", Sonnet 5 Max was going to be off "Agentic Search results" chart. lol ...
In short, Sonnet 5 Low/Medium is more cost efficient, if its a task below Opus 4.8 Medium. For the rest its expensive and your better off using Opus 4.8.
1. They are criminals. Criminals are not bound by laws.
2. Trying to reduce anonymity to go after criminals, simply means giving up anonymity for all but the criminals. See point 1 ... Criminals do not care and will find ways to not get caught.
3. I find the idea of this blanked statement that protesters are criminals insane dangerous and smells of authoritarianism. Peaceful protesters are just that, peaceful. Those that do crimes during protests, are criminals who can be literally caught.
4. The issue of "ability of rioters far exceeds that of the authorities", is more that the authorities do not have their ducks in a row. Blanked mass surveillance is not the solution.
5. Where does it stop? A what point are we running Russia like Max surveillance software on our smartphones, tracking where we go, who we talk too, ... all in the name of catching maybe, some criminals.
> Something need to be done for this.
Its called a better and responsive police force.
> radicalize the youth into wrong paths overnight using social channels
Imaging, that those people who radicalize youth are, ... not using social channel to do so. Wait, ... how did most of the people who ended up going to Syria get radicalized? Most was not via social media, it was with direct contact. Do we ban social contact?
This is just the typical quick fix type of answer. Problem, must be X. No, lets not invest money into police, social councils, case workers, etc...
Thing is, we have seen police getting lazy because, hey, why do investigation work if we can just get free evidence from criminals phones. O, those criminals now encrypted / try to hide data. Ok, so we now need to make it illegal because screw society, we want easier jobs.
No, everybody needs to give up their privacy "for the greater good". You must have something to hide, if you do not let the government read what you wrote, today, yesterday, 10 years ago ...
Have you ever been to China or other countries where saying the wrong thing, can be unpleasant to life changing? Where people learn to not talk what they rally think outside their little family corner. Where corruption is rampant because nobody can protest. Remember, today its your criminal protesters, tomorrow if a government changed into one you do not like, you become the criminal protester.
The right answer is a better funded and accountable police / social structure / help systems. And accountability, to ensure proper policing.
I remember when my grandfather died, and my grandmother got part of his pension.
It put her pension above a max threshold for old age / low pension (cheaper phone bill, no tv taxes, ...) by, ... hold your hats... 2 Euro.
Father called the service ... "sorry, we can not do anything about that".
She lost over 100 Euro in benefits because technically... she was not low pension anymore but was financially WAY more poor then when my grandfather was alive.
So when my grandfather was alive, their combined pension was higher, they qualitied as having a low pension. When only *part* of my grandfather pension got combined into my grandmothers. Well, now your not poor...
I do not remember the exact pensions but it was a massive haircut, with the insulting on top, losing those benefits. The lost benefits made it from threading water, to drowning.
Without my dad financial help, she will have lost the house. Ironically, if she did not get my grandfathers pension part, she will have qualitied for more social assistance, and will have been better off.
The issue is that there are only so many fabs in the world that make memory. And if you want the good stuff, your easily going into 400 ~ 750b parameter models. That means at FP4 400 to 750GB memory.
Did i mention there are only so many memory makers and they are all busy printing money with HBM memory?
Intel is trying with Crescent Island, to make a 160GB GPU that uses LPDDR5X memory.
HBM takes multiple times the resources to make vs basic DDR5 memory. So by going this route, you have more memory, with the disadvantage that its only 700GB/s. VS HBM pumping out Terrabyte numbers like its nothing.
These cards is reasonably priced, may be good alternative to $10k 96GB Nvidia Blackwells... You give up on token generation (heavily memory dependent), for more memory to run larger models at home/office/company servers.
The problem is, again, there are only so many memory makers and its not like the market is flooded with DDR5 memory anymore, as the big 3 moved a lot of production to HBM.
Another approach is Sandisk making HBF ... Flash memory, like your typical NVME but designed around maximum speed. So instead of loading the models into expensive HBM memory, you use the benefits of density in Flash memory, to offload models into that. Cheaper, but slower... But it leaves your expensive HBM memory free for things like KV Cache, Active parameters, etc... So your model will be slower, but your hybrid using it. As in, faster then running a model from system memory with normal DDR memory, but not as fast as HBM.
So yea, there is a lot in development to reduce the dependance of that resource eating HBM memory. For the wafer cost of 1GB HBM, you normally got 4GB normal memory. That is why the world supply of memory dropped. Not just the insane buying but be HBM is just very inefficient in wafer usage.
Can we not use DDR4 production and create some kind of hybrid solution? Sure, but the big 3 moved away from DDR4 in favor of DDR5 a long time ago. We have competition from China with a mix of DDR4/DDR5, but they also need to scale up. Nobody expected to see a large part of the world production vanish into HBM...
Even if its about DDR4 and older nodes, ironically, most companies had been moving away from DDR4. There is only so much wafer capability in the world, to the point that companies are moving to using DDR2 ... Yea, not a typo, like 2007 DDR2! for IOT devices etc, stuff that does not need fast memory. Because even DDR3 got too expensive for them.
Its not like the old nodes are not used anymore ... Like that capacity was sitting idle. It was still in production making other stuff. The only real solution is that we need more fabs, and those take years to build. And the big 3 delayed investing in new fabs for a long time, unsure about the whole AI bubble stuff. Aka, they did not want to make a ton of fabs to end up with over capacity if the AI growth collapsed.
The Chinese government has stopped direct subsidizing solar panels years ago. I think it was around 2019? This resulted in a lot of companies going under at the time.
It did not stop solar panels getting cheaper and cheaper because of the whole integration and mass production (with healthy free market competition).
The last subsidies like export value-added tax rebates for solar panels and lower rebates for batteries are ending in 2027.
China their main power is, the ability to have everything inhouse. Yea, they subsidize a lot of stuff until it hits critical mass, and then you have often a healthy industry with lots of competition.
China alone has like a few 100 car manufactures because of the subsidies, and over time there will be consolidation / buyouts etc but the end result is a healthy new industry that exports. With again, everything internally being produced.
This is why our subsidies fail. We do one sector, often a few companies at best. This results in few competitors, expensive prices, and often reliance on externals that can bankrupt those companies. And que how we wasted again dozens of billions in propping up a industry with no competitive edge.
People can cry about China but they are actually doing work, despite the mass amount of corruption. That is the big difference with here... Mass corruption got in the way of national security, plop, people go to jail. Industry quickly gets their ** together. Here ... give billions, and the money vanishes, with no real consequences.
Yes but unfortunately a lot of the discussion that people participate in, are not done from a corporate point of view, but from a normal consumer level.
And there is a lot of drama in those discussions. GLM 5.2 is a great model for corporations to run, but people only want to hear about running a 35B/27B or maybe a 120B model. And in that market, subscription services are simply way better value for money (take in account the privacy issues).
Everybody wants GPT 5.5/Opus 4.8 Max levels, on a budget that simply is not realistic. And GLM fit in that 4.8 medium/low level.
But then people do not want to be told that running a 750b model in Q2 or Q1 is just going to destroy the models accuracy. And that is still going to cost them 5k+ for that reduced model.
The whole local llm landscape from a consumer point of view, is just filled with odd people. lol.
Corporation really benefit from those models, because spending $90k on a server, is a deductible expense. And they are billed at token prices anyway from all the major providers. So its a even faster ROI on that hardware.
I am surprised that nobody figured out to make a business of selling leftover capacity from corporate llm installations, because there is easily 12h+ just wasted (unless its a large corp that has people in all timezones).
I remember writing a postgresql compatible DB with Opus 4.5, that used S3 as storage and local caching to make it speedy.
Ironically, Opus 4.5 is by todays standards is antique. If you have some knowledge about Databases, it goes a long way.
But you need to do it step by step. Getting the core to work, getting a Pratt parsing going. The whole pgwire protocol ... the data format ...
Step by step ...
With todays Models, your can probably get away with using /goal and telling it to make a postgresql compatible database, while having it run a few days.
Now, making a fun database test project and having it production ready! Big difference!