- cross-posted to:
- [email protected]
- cross-posted to:
- [email protected]
You must log in or # to comment.
It’s becoming painfully obvious that there is no way to ethically use frontier models powered by these monstrosities. It is currently 100 F in Tuckahoe, the largest city in Henrico County… and they’re asking people to not use electricity so that these heat-and-pollution-generating slop factories can use it instead.
This is insanity.
Tuckahoe
They are being paid. Why else would they doom their own people? You won’t stop it if you don’t stop them.
The only reasonable response to these requests is fuck you figure out your data center power issues yourself.
Power is for the people.
Oh, there are definitely ethical ways to use these models. It’s just that those methods are not being enforced by local counties or your governments. Thus, companies are able to do whatever the hell they want, which means it’s going to be unethical by default.
What we need is regulation, enforcement, and a stop letting these companies trample all over everything they want to.
No, there really isn’t. The frontier models are created through massive plagiarism. They’re designed to be addictive to use. They consume massive amounts of resources to feed you slop. They are inherently unethical. We’re burning the planet down to keep them running, and we don’t even have a demonstrable financial ROI to show for it.
Stop using them. If your employer makes you use them, maliciously comply by wasting tokens until the financial pain is too great for them to bear and they stop. If you yourself are addicted, switch to small, local, open-source, open-weight models you can run yourself. You won’t burn the world down running a small model on your own computer.
You have that backwards. The only thing you gain from running local models is privacy. It is not cheaper, it is not more efficient. You are actively hurting the environment MORE by using a local model on your own. LLM efficiency sky rockets the more users there are on a single loaded model.
IMO the only way we get to efficient LLM usage would be by having very efficient non frontier models running only for its local community to use, where you can have assurances on whether its power source is clean or not. That doesn’t help with the plagiarism aspect though
Are you serious?
- Local model: Spends most of its time turned off. Only active when I want it to be active, and only for a little while. Dedicated solely to generating the small amounts of code I use it for. Does nothing else. Costs $0 per token, and electricity costs are negligible.
- Frontier model: Always on, running on millions of GPUs. Would be burning down the planet even if hardly anyone was using it. Incredibly wasteful, being used for trivial tasks and convincing people that their horrible ideas are visionary every day. Misspelling “strawberry” for the masses. Trained specifically to be addictive. Can easily cost a software developer who is addicted to AI thousands of dollars a month, with the recent price increases.
I’d love to see some data to back up the assertion that frontier models are somehow cheaper and more efficient than running a model locally.
You’re probably burning more energy turning it off and on again. It doesn’t really use any noticeable power sitting idle.
Anyway, a direct comparison would be pretty difficult because your model is probably tens of billions of parameters, not over a trillion. Energy consumption per output token will probably be a bit higher for the frontier models but something that people have found is that higher quality models often need fewer tokens to achieve the same goal. Plus how many times do you re-prompt your local model vs Claude Fable or Opus for example to get the desired result?
You’re probably burning more energy turning it off and on again. It doesn’t really use any noticeable power sitting idle.
I am absolutely not burning more energy than a frontier model by doing things like putting my laptop to sleep or shutting down unused services when I want to conserve battery power.
Anyway, a direct comparison would be pretty difficult because your model is probably tens of billions of parameters, not over a trillion.
True.
Energy consumption per output token will probably be a bit higher for the frontier models but something that people have found is that higher quality models often need fewer tokens to achieve the same goal.
That’s actually not true. In fact it’s much the opposite. Frontier models churn through tokens at a much higher rate, because of their higher complexity and higher number of parameters. Research is still new on this, but having a frontier model analyze your code files versus a small, local model for the same task seems to be enormously wasteful. If you must use a frontier model for something, have it do that work after receiving the output from an agent using a small model to read and summarize your code.
Plus how many times do you re-prompt your local model vs Claude Fable or Opus for example to get the desired result?
…Almost never? I’m not a fan of letting AI do much of ANY of my coding, because it will inevitably bloat my codebase with garbage regardless of which model I use. So I severely restrict my model usage to simple, clearly-defined, narrow-scoped tasks that can save me a bit of time, and that’s it. With guardrails and discipline like that, I barely ever have the need to re-prompt.
I am absolutely not burning more energy than a frontier model by doing things like putting my laptop to sleep or shutting down unused services when I want to conserve battery power.
I was under the impression you keep loading the model into VRAM and unloading it when finished using it, I meant it’s less power efficient than just keeping it in VRAM.
That’s actually not true. In fact it’s much the opposite. Frontier models churn through tokens at a much higher rate, because of their higher complexity and higher number of parameters.
Thing is, the input/reading part of it is cheap and wastefully generating extra tokens as output costs you more in energy (or money if using an external service). Put it this way: Claude has historically had 3 models: Haiku (small), Sonnet (medium), Opus (big). Sonnet 5 came out recently and people using Claude Code have reported that it’s so verbose, it’s now more expensive to use for the same task than Opus, which has much bigger costs per Mtok. That would mean it probably also uses more energy than the bigger model.
…Almost never? I’m not a fan of letting AI do much of ANY of my coding, because it will inevitably bloat my codebase with garbage regardless of which model I use. So I severely restrict my model usage to simple, clearly-defined, narrow-scoped tasks that can save me a bit of time, and that’s it. With guardrails and discipline like that, I barely ever have the need to re-prompt.
At that point, why bother with a local model, you could use Deepseek V4 flash and probably spend less than a tenner a month on it. It’s surprisingly capable (I mean sometimes you can barely tell it’s not a frontier model) and costs next to nothing to use.
If you must use a frontier model for something, have it do that work after receiving the output from an agent using a small model to read and summarize your code.
It’s sort of what my workflow does when I use OpenCode. Bigger model (GLM-5.2 or GPT-5.5 depending on which one hasn’t run into its usage limit) reads my prompt, the .md files describing the repo and the overall file structure of the repo, then fires off parallel DeepSeek V4 Flash scouts on usage credits to read and summarize the files as needed. The big model then does the planning and again DeepSeek V4 Flash is the one to execute it via subagents. The subagents running DeepSeek usually come back with 1-2 cents in cost.
I did try a Qwen-3.6 distillation locally and it was pretty capable in terms of output, but it’s more expensive for me than the DeepSeek Flash on API usage costs, since electricity isn’t free here and my GPU is 2 generations old. And it’s slow as hell, since it has to offload a lot to CPU/RAM over GPU/RAM.
The big models I only use as subscriptions that I’m prepared to end at any moment if they reduce the usage I get. Let the AI companies eat the cost, I’ll never pay them API pricing if they want 20 or 30 dollars for a million output tokens.
Very serious. Your personal amount of usage means nothing at all in this conversation. It is entirely about tokens per watt. The amount of energy the memory operations involve scale incredibly well when people are accessing the same object in memory simultaneously. Last I looked it was around a 10x difference for the same models efficiency.
If you want me to be your personal search engine you’ll need to wait a bit, im making dinner right now and would rather look for the articles on my desktop.
Very serious. Your personal amount of usage means nothing at all in this conversation. It is entirely about tokens per watt. The amount of energy the memory operations involve scale incredibly well when people are accessing the same object in memory simultaneously. Last I looked it was around a 10x difference for the same models efficiency.
Hold up. Are you talking about caching? Because if you are… yeah. That has nothing to do with the model and everything to do with the service layer around the model. The same service layers can be - and have been - implemented in tools like Lemonade Server, llama.cpp, Ollama, etc.
And I really do want to know your sources.
Mine say GPT 5.5 is probably using quite a lot more than 0.34 Wh per query (0.34 Wh is what Sam Altman claimed for the then-current version of GPT in June of 2025, but he hasn’t released numbers since then and no one has done an independent analysis). With Claude, an independent estimate from last year pegged Sonnet at 0.8 Wh for a short prompt, 2.8 Wh for a medium one, and 5.5 Wh for a long one. Current numbers are, again, almost certainly much higher. And just for fun, there’s DeepSeek (which I’ve never used and never would use), with the reasoning-tuned DeepSeek-R1 hitting a whopping 29 Wh for a complex query.
Meanwhile, small, open models are probably in the 0.07 - 0.2 range, depending on the model, the hardware it’s running on, and the nature of the query. Of course, there are much weightier open models too, with ones like Llama 3.1 405B using about 9 Wh for a medium-length prompt. On the other hand… who is going to run that on their local machine?
Look… If I’m wrong, and using local models the way I do - sparingly and infrequently - really does consume more electricity than using Claude Code, I want to know. I have no problem whatsoever with eschewing AI models entirely, since I despise all of them. But given how tight-lipped OpenAI and Anthropic are about energy consumption per average prompt, and what independent analyses have estimated, I am highly skeptical that they are acting as some sort of paragons of environmental stewardship.
Not talking about caching (though there would be some decent memory savings due to that on general platforms like ChatGPT and tools like Codex). I am talking about large batch sizes, which are concurrent requests all accessing the same memory at the same time. The model is loaded once onto the GPU(s) and then many simultaneous requests can read that memory at the same time. When those requests are all processing their responses simultaneously, the energy per token drops off a cliff.
And yes, running a smaller model would generally take less power, but thats not really a fair comparison. Small models just wont give you the same results as larger ones. You need to compare it apples to apples. If you want to compare your local Qwen model running on your laptop, you compare those numbers to larger systems supplying that same qwen model to thousands of people. Just because we are comparing cloud services to local doesn’t automatically mean GPT 5.6 vs Qwen 3.6 27B. There are plenty of cloud AI providers running all sorts of models and sizes.
As for one of the articles I learned alot of this from originally, this is one I recommend going through. It really goes deep into the whole topic: https://arxiv.org/html/2601.22076v1
Won’t somebody think of the datacenters??!
Won’t somone think of the job creators??!
Won’t somebody think of the billionaire fossil fuel owners?
What about the shareholders?
They should be investigating more wisely?
If Virginians aren’t surrounding the county government building, 24/7, with firearms…
Your fucking state motto is Sic Semper Tyranis.
Live up to it, or never be taken seriously, about anything, ever again.
You know what happened in Iceland in the 07/08 financial crisis?
Something like 1/3 to 2/3 of the entire population of the island surrounded parliament, literally with pitchforks, untill they actually prosecuted the criminal financiers for their actual financial crimes.
Iceland is basically the only Western country that didn’t bail out their bankster fraud artists, and sent them to fucking prison, instead.
Ya’ll gonna let your selves get outdone by Iceland?
Ya’ll gonna let your selves get outdone by Iceland?
I’ve seen, firsthand, how Icelanders group up and solve problems together as a unit. They have a cultural tradition of outdoing other cultures, because of this.
The state of Virginia cries about the battle flag the 1rst Minnesota took during the Battle of Gettysburg from the 28th Virginia on the July 3 day of the battle to this day.
They have asked for the flag back 4 times. Each time we say No!. The 1rst Minnesota took that flag in open battle and the 70% casualties was the blood paid for it. The blood of 4 men directly paid for its capture.
Their state motto hasn’t meant much since that battle.
Hah!
Well shit, I was not aware of that.
… that’s uh, funny, I guess, in a more modern context, with more temporal distance.
But, Minnesotans are bleeding again lately, due to armed tyrants, so I guess in that sense, its less funny.
Iceland has 400,000 people, and is an island. America has 340 MILLION people, and is 3000 miles wide.
In America, it’s a LOT more difficult to gather a large population in one place. We tend to get gatherings in multiple cities, instead. The last big one had 3,300 hundred different protests around the nation, with over 7 million people.
Iceland has 400,000 people, and is an island. America has 340 MILLION people, and is 3000 miles wide.
Lets reread the first line of the post you quoted:
If VIRGINIANS aren’t surrounding the COUNTY government building, 24/7, with firearms…
They are talking about a county in a state. Not America.
If you actually read the article, nobody is asking individuals to reduce usage, the local government is trying to save on its own bill by… Turning off unused lights and computer monitors at work.
The real crime is variable rate power plans and price hikes. Shouldn’t exist. It’s a utility, why does it need a market price? Should be provided by government, not private companies…
That they are in the position where they even need to ask is an admission that they are sell out traitors to the inherent responsibilities and duties vested in them by the voting public.
They are frauds and failures, the only correct thing they could all do is rescind all permits for any data center contracts, and then resign immediately.
Anything less than that?
Run them out of town, outta the entire fucking state.
If I lived in Virginia County, I’d do my part and use as much electricity as I could. Bring the grid down, piss everyone off, and call out the fucktards who are blaming “the people”, rather than the data centers.
Why couldn’t they ask data centers to conserve power?
But how would I ask Claude the best way to cook beans at 2000ft above sea level?
At 2000 ft? Just cook them. I don’t think you really need to make adjustments until you’re about 3x higher in elevation than that…
Thanks, beans on the menu tonight now
Oh, so Virginia is going to be the sacrificial state, good for them.
They voted for all this shit.
I believe VA has more data centers than any other state.
Why are residential and critical infrastructure electric rates uncapped? Prices should not be allowed to be hiked. The material ability to produce the electricity is there no matter what and demand is functionally inelastic–this price hiking doesn’t cover any increase in cost of production, it just goes directly to utility company profit lines. We need to stop fucking negotiating with power companies.
I wouldn’t point to power companies as the sole culprits here. These data centers shouldn’t be getting approval in the first place based on their power and water demand, but they promise more money and jobs coming to the area and these dipshits keep believing them.
So long as fuel prices fluctuate, power prices will fluctuate. Some price variation is more for curbing costs than for padding profits. I’m not the guy running those numbers at my power plant, but I see power rates swing with demand and prices/MW. The past few days in VA were pretty demanding on us.
I think there’s plenty blame to be laid at the feet of data center owners and local permitting morons, for sure. That said, at the end of the day, National Grid NA has posted year on year record profits and continues to argue that price hikes are necessary due to short term fuel cost spikes. Meanwhile, regulators and legislators insist that we need to “negotiate” pricing when most of the infrastructure and labor for generation and transmission is subsidized by the state, is a critical infrastructure, and the operators are still fully capable of netting immense profit despite everything. It just doesn’t make sense. They are not going to go away because their profits are fewer billions this year than last year’s billions, and if they did, it would be more than fine for the state to step in and take over in providing critical infrastructure for the operation of society, even if it is a cost center, because it’s one of the key fucking reasons humans collect into cooperative society in the first place.
100% reasonable and fair take. No notes lol.
I don’t know who needs to hear this, but NEVER fly a drone near a transformer
Hahahahahah
all you be hearing is datacenter noise.
Especially with a long steel wire dangling from it.
That sucks. I remember everyone pointing and laughing at California because the state was being defrauded by Enron, which caused regular brownouts.
I ain’t do that. If you are in Virginia (or anywhere else, but since this post is about Virginia) and need help cooling off, pour water on your forearm. I learned it as left forearm because it’s barely closer to your heart, but I haven’t noticed a difference. You’re getting better air cooling because of the water, and it’s on your forearm which gets a lot of bloodflow. Also you aren’t getting your clothes wet. Old farmer’s trick.
Mentioning Enron never ceases to make my blood boil. People died for market manipulation. Arguably the best public utility in the US was cannibalized for corporate profits.
The reason Texas didn’t bother connecting to the grid was because it had a higher uptime than the rest of the nation. It was literally, the energy capital of the world.
They privatized the public utility and within a year Texas went from the cheapest electric in the nation to the most expensive.
Worst of all, no one at Enron got the chair for their murders. In fact, because it was based in Texas when things went to shit and the execs took their golden parachutes. A judge ruled that no return of the stolen assets was required because one of the executives died. Imagine robbing a bank, giving the money to your wife, and then dying and your wife gets to keep the money.
Fuck those stupid fucking conservatives and the republicans representing them. Ultimately, the tolerance of their corruption and greed has brought the USA to its knees.
one of the worst parts was all the accounting students thinking “wow! i can do that someday too!”
I lived through those brownouts but I was a teen and didn’t understand what was going on, just that it was annoying. We had just moved into a house in a very hot part of town where the wind goes to die and it suuuuucked.
Fuck that. Now I am going to shower even harder.
At your local government job? Bold move
This is off topic, but toms hardware is one of a growing list of pages using “ad recovery” services that intentionally break their own page and then blame DNS-level filters for the problem. It’s disingenuous at best and I really don’t think a site willing to employ a such deceptive tactic should be allowed to get page views.
sure would be a shame if someone hooked up the 4X4 to the power line and knocked it down. or got a back hoe and dug it up… just saying thats a great way to conseve power. i hope to hear about the AI outage black outs all next week…
