ChatGPT 5 power consumption could be as much as eight times higher than GPT 4 — research institute estimates medium-sized GPT-5 response can consume up to 40 watt-hours of electricity

Comments

• Sam_Bass@lemmy.world ⁨7⁩ ⁨months⁩ ago

  And how many milliwatts does an actual brain use?

  source

  • homesweethomeMrL@lemmy.world ⁨7⁩ ⁨months⁩ ago

    1.21 jigawatts

    source

    • Sam_Bass@lemmy.world ⁨7⁩ ⁨months⁩ ago

      Ahh but you need 8 for a round trip

      source
  • Warl0k3@lemmy.world ⁨7⁩ ⁨months⁩ ago

    It’s shockingly tricky to answer precisely, but the commonly held value is that a human brain runs on about 20w, regardless of the computational load placed on it.

    source

    • Sam_Bass@lemmy.world ⁨7⁩ ⁨months⁩ ago

      So maybe .5 kwh/d. Still vastly more efficient than a code-based one.

      source
• skisnow@lemmy.ca ⁨7⁩ ⁨months⁩ ago

  There’s such a huge gap between what I read about GPT-5 online, versus the overwhelmingly disappointing results I get from it for both coding and general questions.

  I’m beginning to think we’re in the end stages of Dead Internet, where basically nothing you see online has any connection to reality.

  source

  • ipkpjersi@lemmy.ml ⁨7⁩ ⁨months⁩ ago

    Well yeah, it’s a for-profit company. They exist solely to make money, that’s their entire goal.

    It’s almost all marketing and has been for a while. ChatGPT peaked with 4o (and 4.5 if you used their API), 4.1 was a step backwards despite them calling it an improvement, and 5 was another step backwards.

    source
  • ghen@sh.itjust.works ⁨7⁩ ⁨months⁩ ago

    The stock market is barely connected to reality and that is required to be updated every 3 months by every single company. Just imagine what the internet’s going to be like.

    source
  • CheeseNoodle@lemmy.world ⁨7⁩ ⁨months⁩ ago

    People who fawn over generative AI haven’t tried to use it for more than 5 seconds. I wish it could run a ttrpg game for me or even just remember the details of its original prompt but its not even close.

    source
• AgentOrangesicle@lemmy.world ⁨7⁩ ⁨months⁩ ago

  Isn’t this the back plot of the game, Rain World? With the slug cats and the depressed robots stuck on a decaying world when the sapient, organic species all left?

  source

  • Hadriscus@jlai.lu ⁨7⁩ ⁨months⁩ ago

    I didn’t know there was such a backstory

    source
  • Patches@ttrpg.network ⁨7⁩ ⁨months⁩ ago

    Spoilers dude.

    source
• antihumanitarian@lemmy.world ⁨7⁩ ⁨months⁩ ago

  The last 6 to 12 months of open models has pretty clearly shown you can substantially better results with the same model size or the same results with smaller model size. Eg Llama 3. 1 405B being basically equal to Llama 3.3 70B or R1-0528 being substantially better than R1. The little information available about GPT 5 suggests it uses mixture of experts and dynamic routing to different models, both of which can reduce computation cost dramatically. Additionally, simplifying the model catalogue from 9ish(?) to 3, when combined with their enormous traffic, will mean higher utilization of batch runs. Fuller batches run more efficiently on a per query basis.

  Basically they can’t know for sure.

  source
• jsomae@lemmy.ml ⁨7⁩ ⁨months⁩ ago

  For reference, this roughly equivalent to playing a PS5 game for 4 minutes (their estimate) to 10 minutes (their upper bound)

  calulation

  source www.ecoenergygeek.com/ps5-power-consumption/ PS5 standard usage: 200 watts ± 100; call it 200 W TV: 27 watts - 134 watts → call it 60 W URI’s estimate: 18 Wh / 260 W → 4 minutes URI’s upper bound: 48 / 260 W →10 minutes

  source

  • bier@feddit.nl ⁨7⁩ ⁨months⁩ ago

    It is also the equivalent of letting a LED light bulb run for an entire day (depending on bright it is, some LED bulbs use under 2 watts of power).

    source
  • MangoCats@feddit.it ⁨7⁩ ⁨months⁩ ago

    I was just thinking, in more affordable electric regions of the US that’s about $5 worth of electricity. You’d tip a concierge $5 for most answers you get from Chat GPT (if they could provide them…) and the concierge is likely going to use that $5 to buy a gallon and a half of gasoline, which generates a whole lot more CO2 than the nuclear / hydro / solar mixed electrical generation, in reasonably priced electric regions of the US…

    source

    • jsomae@lemmy.ml ⁨7⁩ ⁨months⁩ ago

      That doesn’t seem right. By my calculations it should be like 5¢.

      source

      • -> View More Comments
  • buttnugget@lemmy.world ⁨7⁩ ⁨months⁩ ago

    I love playing PS5 games!

    source
• plyth@feddit.org ⁨7⁩ ⁨months⁩ ago

  That are 25 request per kwh. At 10 to 25cents per kwh that’s 1cent per request.

  source

  • gens@programming.dev ⁨7⁩ ⁨months⁩ ago

    Hmmm. Sure. But I find people don’t understand how much one kWh really is. A 500W drill can twist your arm. Imagine yourself twisting someones arm with all you got for a whole hour. Or idk. Either way it’s a lot of energy.

    And then you think about how much more energy a car uses then a human does. And then you find out about hot water…

    source
  • MangoCats@feddit.it ⁨7⁩ ⁨months⁩ ago

    Which is why they’re giving everybody free access, for now.

    source
• RememberTheApollo_@lemmy.world ⁨7⁩ ⁨months⁩ ago

  Help me out here. What designates the “response” type? Someone asking it to make a picture? Write a 20 page paper? Code a small app?

  source

  • ckmnstr@lemmy.world ⁨7⁩ ⁨months⁩ ago

    Response Type is decided by ChatGPTs new routing function based on your input. So yeah. Asking it to “think long and hard”, which I have seen people advocating for to get better results recently, will trigger the thinking model and waste more resources.

    source

    • towerful@programming.dev ⁨7⁩ ⁨months⁩ ago

      So instead of just saying “thank you” I now have to say “think long and hard about how much this means to me”?

      source

      • -> View More Comments
• Deflated0ne@lemmy.world ⁨7⁩ ⁨months⁩ ago

  And an LLM that you could run local on a flash drive will do most of what it can do.

  source

  • lefixxx@lemmy.world ⁨7⁩ ⁨months⁩ ago

    Can you give an example?

    source
  • EncryptKeeper@lemmy.world ⁨7⁩ ⁨months⁩ ago

    I mean no not at all, but local LLMs are a less energy reckless way to use AI

    source

    • Corkyskog@sh.itjust.works ⁨7⁩ ⁨months⁩ ago

      Why not for the ignorant, such as myself?

      source

      • -> View More Comments
  • ckmnstr@lemmy.world ⁨7⁩ ⁨months⁩ ago

    Probably not a flash drive but you can get decent mileage out of 7b models that run on any old laptop for tasks like text generation, shortening or summarizing.

    source
  • Tikiporch@lemmy.world ⁨7⁩ ⁨months⁩ ago

    What do you use your usb drive llm for?

    source

    • outhouseperilous@lemmy.dbzer0.com ⁨7⁩ ⁨months⁩ ago

      Porn. Obviously.

      source
• stevedice@sh.itjust.works ⁨7⁩ ⁨months⁩ ago

  The team measured GPT-5’s power consumption by combining two key factors: how long the model took to respond to a given request, and the estimated average power draw of the hardware [they believe is] running it.

  source

  • T156@lemmy.world ⁨7⁩ ⁨months⁩ ago

    Basically, it’s just a number whipped out the backside, since response delay varies a lot, and I don’t think we know what hardware OpenAI uses exactly.

    source
• sp3ctr4l@lemmy.dbzer0.com ⁨7⁩ ⁨months⁩ ago

  Fucking Doc Brown could power a goddamn time machine with this many jiggawatts, fuck I hate being stuck in this timeline.

  source
• Dasus@lemmy.world ⁨7⁩ ⁨months⁩ ago

  that’s a lot. remember to add “-noai” to your google searches.

  source

  • Opisek@lemmy.world ⁨7⁩ ⁨months⁩ ago

    This is my weekly time to tell lemmings about Kagi, the search engine that does not shove LLM in your face (but still let’s you use it when you explicitly want it) and that you pay for with your money, not your data.

    source
  • macaw_dean_settle@lemmy.world ⁨7⁩ ⁨months⁩ ago

    Or just use any other better search like Bing or duckduckgo. googol sucks and was never any good. Quit pushing ignorant garbage.

    source

    • homesweethomeMrL@lemmy.world ⁨7⁩ ⁨months⁩ ago

      googol sucks and was never any good.

      Ha! Kids these days.

      source
    • Dasus@lemmy.world ⁨7⁩ ⁨months⁩ ago

      duckduckgo yes, but … bing?

      source

      • -> View More Comments
  • Ilovethebomb@sh.itjust.works ⁨7⁩ ⁨months⁩ ago

    I’m just going to ignore the AI recommendations, let them burn money.

    source

    • Dasus@lemmy.world ⁨7⁩ ⁨months⁩ ago

      i don’t judge you for that. honestly letting it matters fuck all at this point

      source
• pHr34kY@lemmy.world ⁨7⁩ ⁨months⁩ ago

  Tech hasn’t improved that much in the last in the last decade. All that’s happened is that more cores have been added. The single-thread speed of a CPU is stagnant.

  My home PC consumes more power than my Pentium 3 consumed 25 years ago. All efficiency gains are lost to scaling for more processing power. All improvements in processing power are lost to shitty, bloated code.

  We don’t have the tech for AI. We’re just scaling up to the electrical senand demand of a small country and pretending we have the tech for AI.

  source

  • Caitlyynn@lemmy.blahaj.zone ⁨7⁩ ⁨months⁩ ago

    Not even the ai tech itself is enough for ai

    source
  • SCmSTR@lemmy.blahaj.zone ⁨7⁩ ⁨months⁩ ago

    It’s the muscle car era: can’t make things more efficient to compete with Asia? MAKE IT BIGGER AND CONSUME MORE

    source
  • ChokingHazard@lemmy.world ⁨7⁩ ⁨months⁩ ago

    This is nonsense, an M1 runs many multiples faster and at much lower wattage.

    source
• Blackmist@feddit.uk ⁨7⁩ ⁨months⁩ ago

  That’s alright. When they’ve got a generation of people who can’t even hold a conversation without it, let alone do a job, that price increase will drop that energy use pretty rapidly.

  source
• TheGrandNagus@lemmy.world ⁨7⁩ ⁨months⁩ ago

  I have an extreme dislike for OpenAI, Altman, and people like him, but the reasoning behind this article is just stuff some guy has pulled from his backside.

  source

  • gil2455526@lemmy.eco.br ⁨7⁩ ⁨months⁩ ago

    Yeah, I saw some people speculating GPT-5 is having worse performance in some tasks than older models but still being forced to everyone because it’s a consolidation and cost cutting version, but how could it be cutting costs if it consumes 8 times more power?

    source
  • MangoCats@feddit.it ⁨7⁩ ⁨months⁩ ago

    This figure is already not bad. 40 watt hours = 0.04kWh - you know kWh? That unit on your electric bill that is around $0.18 per kWh (and data centers tend to be in lower cost electric areas, closer to $0.11/kWh.) Still, 40Wh would register on your home electric bill at $0.0072, less than a penny. For comparison, an average suburban 4 ton AC unit draws 4kW per hour - that 40Wh request? 1/100th of an hour of AC for your home, about 36 seconds of air conditioning. I don’t know that this article is making anybody “look bad” in terms of power used.

    source

    • msage@programming.dev ⁨7⁩ ⁨months⁩ ago

      What exactly do you get for that power though?

      The point is that it’s too much power for little gain in return.

      source

      • -> View More Comments
  • outhouseperilous@lemmy.dbzer0.com ⁨7⁩ ⁨months⁩ ago

    Thats actyally a fav rhetorical trick of mine when arhuing with consummatw bullshitters who have followers.

    source
• DarkCloud@lemmy.world ⁨7⁩ ⁨months⁩ ago

  This bubble needs to pop, the sooner the better.

  source
• Melonpoly@lemmy.world ⁨7⁩ ⁨months⁩ ago

  It takes less energy to dry a full load of clothes

  source

  • morto@piefed.social ⁨7⁩ ⁨months⁩ ago

    Maybe you're mixing Wh with kWh, but still, 40Wh is a lot for a single request

    source

    • Ilovethebomb@sh.itjust.works ⁨7⁩ ⁨months⁩ ago

      Roughly the capacity of a laptop battery, a huge amount of energy per request.

      source

      • -> View More Comments
    • Melonpoly@lemmy.world ⁨7⁩ ⁨months⁩ ago

      Yeah I think I have

      source
  • MangoPenguin@lemmy.blahaj.zone ⁨7⁩ ⁨months⁩ ago

    A standard dryer is more like 2-5kWh for a load… Far more than 40Wh.

    source
  • LodeMike@lemmy.today ⁨7⁩ ⁨months⁩ ago

    40 watt-hours? That’s the energy usage of a very small laptop.

    source

    • Dremor@lemmy.world ⁨7⁩ ⁨months⁩ ago

      Imagine if you had to empty your whole lnaptop battery every time you had to generate a 20 response that may not even be correct… That’ll end up consuming power really fast.

      source
    • jj4211@lemmy.world ⁨7⁩ ⁨months⁩ ago

      Well over the course of an hot or two, but it’s correct that a dryer run even with heat pump is significantly more than 40wh

      source

      • -> View More Comments
• Steve@startrek.website ⁨7⁩ ⁨months⁩ ago

  And it sucks even worse.

  source
• kescusay@lemmy.world ⁨7⁩ ⁨months⁩ ago

  How the hell are they going to sustain the expense to power that? Setting aside the environmental catastrophe that this kind of “AI” entails, they’re just not very profitable.

  source

  • homesweethomeMrL@lemmy.world ⁨7⁩ ⁨months⁩ ago

    Not just”not profitable”, they don’t make any money at all. Loss only.

    source
• yesman@lemmy.world ⁨7⁩ ⁨months⁩ ago

  I think AI power usage has an upside. No amount of hype can pay the light bill.

  AI is either going to be the most valuable tech in history, or it’s going to be a giant pile of ash that used to be VC capital.

  source
• Boxscape@lemmy.sdf.org ⁨7⁩ ⁨months⁩ ago

  Image

  source
• DrFistington@lemmy.world ⁨7⁩ ⁨months⁩ ago

  But we get a huge increase in accuracy, from 30% to 30.5%! And it only took 5x the energy consumption!

  source
• A_norny_mousse@feddit.org ⁨7⁩ ⁨months⁩ ago

  I don’t care how rough the estimate is, LLMs are using insane amounts of power, and the message I’m getting here is that the newest incarnation uses even more.

  BTW a lot of it seems to be just inefficient coding as Deepseek has shown.

  source
• brucethemoose@lemmy.world ⁨7⁩ ⁨months⁩ ago

  I don’t buy the research paper at all. Of course we have no idea what OpenAI does, but Deepseek’s publish papers suggest it’s much more than 1 model per node… I think they recommended like a 576 GPU cluster total.

  That, and going by the really small active parameter count of gpt-oss, I bet the model is sparse as heck.

  There’s no way the effective batch size is 8, it has to be waaay higher than that.

  source
• Valmond@lemmy.world ⁨7⁩ ⁨months⁩ ago

  40Wh or 18Wh which is it?

  That’s my old gaming PC running a game for 2min42sec-6minutes … Roughly.

  source
• vegeta@lemmy.world ⁨7⁩ ⁨months⁩ ago

  Image

  source
• nightwatch_admin@feddit.nl ⁨7⁩ ⁨months⁩ ago

  Of course there are comments doubting the accuracy, which by itself is valid, but they are merely doing it to defend AI. IMHO, even at a fifth of the estimates, we’re talking humongous amounts of power, all for a so-so search engine, half arsed chatbots and dubious nsfw images mostly. And let’s not forget: it may be inaccurate and estimates are TOO LOW. Now wouldn’t that be fun?

  source
• eager_eagle@lemmy.world ⁨7⁩ ⁨months⁩ ago

  Bit of a clickbait. We can’t really say it without more info.

  But it’s important to point out that the lab’s test methodology is far from ideal.

  The team measured GPT-5’s power consumption by combining two key factors: how long the model took to respond to a given request, and the estimated average power draw of the hardware running it.

  What we do know is that the price went down. So this could be a strong indication the model is, in fact, more energy efficient. At least a stronger indicator than response time.

  source