The latest deepseek v4 pro model is 2-5x cheaper than Claude Sonnet 4.6. Cursor's Compose 2.5 that was just recently released is 6x cheaper than Sonnet.
The state of the art models are going to get better and more expensive and smaller models are going to get cheaper.
There will be a point where the intelligence of both the cheap and state of the art models are indistinguishable by humans like it is indistinguishable for me to understand the difference the difference between Terrance Tao and my university math professor.
I don't always need the smartest and most expensive models. I will need it every once in awhile and will gladly pay that price if I had to. What I do need is the model that will solve the current problem I have in a reasonable amount of time.
I know it comes off as pedantic to point this out but: Those are open weight models not open source models.
Closed weight models are the equivalent of SaaS. Open weight models are the equivalent of binary driver blobs or Windows software. We don't really have actual open source LLMs, which would need to publicly release their training data and technique so you could train a similar model yourself, or use their work as a baseline for your own model.
This distinction matters because an actual open source LLM would be extremely important from an ecosystem point of view, if someone ever actually released one.
There are absolutely fully open source models. These are not frontier models, but they very much do exist. OLMo is one of the models explicitly mentioned as having passed the OSI's validation phase. Pythia was also validated by the OSI as meeting its requirements for an open-source AI system. Lucie-7B is a multilingual model is one of the first LLM compliant with the OSI AI definition. Its creators explicitly state that the training dataset, data preparation code, and model weights are all publicly available under open licenses.
>The latest deepseek v4 pro model is 2-5x cheaper than Claude Sonnet 4.6. Cursor's Compose 2.5 that was just recently released is 6x cheaper than Sonnet.
It's ironic how in a thread about "AI subsidies" that people don't think free model releases from AI don't count as subsidies. Whatever AI winter that would cause AI companies to stop subsiding tokens, would probably cause other AI labs to stop doing free model releases. They might not be able to un-release the current crop of open models, but assuming proprietary model development still happens, they'll quickly go obsolete.
The currently-released models don't really go away. Even if they collectively only release a new model every few years for the sake of influence and public image, that's plenty enough to keep the competitive aspect going.
>Even if they collectively only release a new model every few years for the sake of influence and public image, that's plenty enough to keep the competitive aspect going.
This is unpersuasive. Why would AI companies (American or Chinese) stop subsidizing tokens, but keep doing open model releases? At least for the former you can argue it's a lead generation tool for enterprise contracts (eg. hobbyist uses claude code personal plan, then asks the company to buy claude code enterprise, which are billed at API rates), but what's the business case for doing open model releases? You might get some mindshare, but are also arming your competitors in the process. Moreover what makes you think the model releases will be at all competitive to frontier models? Google released gemma 4 a few weeks ago to acclaim, but it's in no way competitive to even GPT-5.4 or Opus 4.6.
> The state of the art models are going to get better and more expensive and smaller models are going to get cheaper.
Why do you think this will be true?
Right now I see the major US labs betting on gaining an advantage from having way more compute, and I see Chinese labs competing with one another in a resource-scarce environment, so they place much more emphasis on compute-efficiency.
But the supply chains that feed into the massive data center growth in the US are strained; there are energy, memory, and logistical bottlenecks to name a few.
In the medium-long run, compute capacity will not grow exponentially forever. Somehow it has for decades, but there can be no infinite exponential growth, and that point may be when the planet really starts to cook itself.
Maybe the US labs will become more compute-constrained, and then have to compete on efficiency.
Or maybe things change fundamentally in some other way I'm not thinking of.
The labs have a perverse incentive to make things as expensive compute wise as possible. The only thing keeping this somewhat in check is competition, but it's intentionally being gatekept by locking up the supply of computing infrastructure. With 3 players it's pretty easy to collude even if indirectly. They can't burn trillions forever. Nvidia's 75% profit margins are not sustainable forever.
>The labs have a perverse incentive to make things as expensive compute wise as possible. The only thing keeping this somewhat in check is competition, but it's intentionally being gatekept by locking up the supply of computing infrastructure. With 3 players it's pretty easy to collude even if indirectly.
By all accounts the AI capex boom is justified up by actual usage, rather than some nefarious plan for "locking up the supply of computing infrastructure". Just look at people complaining about claude availability and anthropic adding various load-shedding measures a few months ago.
sorry to nitpick (I totally agree with what ur saying btw, I run Ministral-3b on my hardware as my go-to bc I don't usually need the "smartest and most expensive models")
>The latest deepseek v4 pro model is 2-5x cheaper than Claude Sonnet 4.6. Cursor's Compose 2.5 that was just recently released is 6x cheaper than Sonnet.
The only way you're running Deepseek V4 with comparable quality/performance is through OpenRouter, at which point you're still susceptible to being price gouged in the future, or by spending >$20k on hardware.
> Anthropic’s CFO testified under oath this March that the company spent $10 billion on compute and made $5 billion in revenue (Ed Zitron has the math). The labs are underwater on inference. They’re raising prices to keep the lights on.
'The labs are underwater on inference' is an absurd thing to say whilst not separating the cost of _compute_ out into training and inference.
According to Dario Amodei, Anthropic are even profitable when including inference as long as you look at it on a per-model basis; it’s just that every model is more expensive to train than the last one.
For instance, if you have already spent $n to train a model and are currently earning $2n selling inference with it; but are concurrently spending $3n training the next model in anticipation of earning $6n with it, then you are already in the hole for $n and are currently also losing $n – but you are doubling your money with each model because your $n investment in the first model returns $2n and your $3n investment in the second model returns $6n.
How is training vs inference any different than other product spaces, where all the costs of bringing a product to market have to be considered for profitability? You can't just look at marginal production cost. You are still underwater if the other development costs are not being recouped by the final sales revenue.
The whole commercial AI enterprise is not economically viable if the inference revenue will not cover both inference and the amortized training costs. Given how fast they are churning through models to compete, you cannot act like the training is an asymptotically low cost.
I wonder how much of Uber blowing their AI budget and MSFT pulling their claude code licenses can be attributed to "tokenmaxxing".
When Meta announced token leaderboards and other followed, I could see this being the logical conclusion. That whole trend is so dumb because it leads to this.
Company announces they will measure developer performance by how many tokens they burn and constantly talks about how the best developers burn the most tokens. Developers see the message and start burning tokens. And then the company acts surprised when their bills go through the roof.
I personally use my OpenAI subscription pretty heavily, 2-3 agents running practically all day on various tasks but I never even get close to running into limits while I hear about others blowing through limits on multiple accounts in the same time period. I'm convinced that most of those folks and their elaborate workflows aren't really for productivity but for bragging rights about how much they use AI.
The same here, where I haven't come close to hitting any of my CC limits. Even though I'm more productive than I've ever been (as measured by finished, valuable tasks running in production) and I'm clearing out months of backlog, I have either one of two conclusions when I hear about others who suggest they need more:
1. I'm doing it wrong. Apparently I'm supposed to give it a vague paragraph about what the business does, and I can run off and sip margaritas and wake up to a fully fleshed business
2. They don't know what they're doing, and they're sending the LLM off on a wild goose chase that it does a reasonable job of working it's way out of, so they consider it success despite the waste.
> I personally use my OpenAI subscription pretty heavily, 2-3 agents running practically all day on various tasks but I never even get close to running into limits
Same. But if I was working for an organization that measured token usage, you can bet I would be doing things like creating a cron job that uses claude to create a customized bespoke report update of the current status of all my open assigned tickets and message that to myself 4 times a day... token burn for zero purpose whatsoever.
> I'm convinced that most of those folks and their elaborate workflows aren't really for productivity but for bragging rights about how much they use AI.
This is quite the reductive, charged statement. Can I ask what subscription plan you're using?
My personal experience is unlike this at all-- I work on ever-expanding codebases so I can easily burn tokens. Not to mention, structured agentic coding with adverserial reviews & task organization is not token-efficient. Additionally, for the problems I'm working on, only xhigh or high reasoning gives me worthwhile results while saving time. There are definitely configurations where default consumption doesn't work.
For reference, I used 15 billion tokens (most of it cached) last month on my day job's enterprise plan. That doesn't include my personal plans' usage.
I make like 2 prompts a week to gemini flash on the weband get more done than all the people that are exhibiting literal manic behavior in the way they use LLMs.
I really wish the management behind these dumb ideas couldnt just quietly pretend they never did it once it goes out of fashion.
The fact that somebody established a leaderboard for tokenmaxxing ought to follow you around like a black cloud for the rest of your career once the collective hallucination lifts and people realize just how monumentally stupid it was.
Alas they do all these stupid things together which makes it seem more defensible and then everybody forgets.
In the three options OP presents, I wonder if there's a fourth: BYO model
Customers give vendors metered access to their model. They can budget tokens per vendor. Vendors selling "AI products" can have a cleaner story and win on the margin.
The first step to is to iron out a reasonable protocol, basically authorizing a, access token, and then the model providers (OpenAI, Anthropic, etc.) do the rate limiting. Theoretically this could be done by OpenRouter too.
But even so - do customers want an "AI product" packaged cleanly, or do they want to manage token capacity? They may be forced to do the latter....
Not mentioned in the article/blog was the local alternative. Many applications will run just fine locally and not in the cloud. This is also more secure. Running local will probably eventually become the norm. It makes me wonder about the future of all these VC funded AI companies...
I seldom use my PC anymore ever since i got a laptop. with the cost per token increasing along with the random "features" where models will just eat through your tokens in one hour. I really have been tempted to turn my PC into a server to run local models on there
Insofar as I can tell, inference is on a certain path toward becoming "free". The models are now extremely powerful on high-end consumer hardware, and the efficiency trend seems likely to continue.
Here is a recent non-rigorous benchmark I ran against a bunch of models. Qwen3.6 35B A3B fine-tuned with opus data runs plenty fast on my local machine and produce outstanding results - easily in the top 5, comparable to GPT 5.5 Pro (which is $180/mtok).
I've predicted for years now that the industry will head down the path of the virus scanning vendors: selling subscriptions to be able to download the latest versions of models. I simply don't see how any other business model is remotely viable, except at the very highest end of inference or video gen.
Has this not been true for a long time now? Most companies have had enterprise/business level prices that was highly connected to usage for a what feels like at least a year.
This is only true if your world is limited to openai, antropic and alike.
There are a whole bunch of companies somewhere else in the world that are getting better and cheaper every month, hardware side included. all without the infinite VC money
My expectation is that local models will be the default for coding within a year or two. You can already run Qwen 3.6 with MTP at a pretty reasonable speed without needing a huge amount of VRAM. And while it's not as good as current frontier models, it's already quite competent for a lot of tasks.
And there's no sign that people are running out of ideas for how to optimize models further. You see a bunch of papers come out literally every few weeks right now. So, it's entirely plausible to me that we'll see models that are superior to current frontier ones in a year or two that will run on your machine.
Once we get to that point, I don't think it's even going to matter if frontier models keep improving for most people. Being able to run the model on your machine, use it as much as you want in any way you want, without having to worry about it changing from under you or the company changing pricing, and not have to send all your data to the vendor are going to be the deciding factors.
At some point the models are just good enough to do what you need to do. On top of that, I expect tooling around models and coding patterns will evolve as well. That could compensate significantly for the capabilities of the model. We already see this happening with two prime examples here:
Some of these coming price increases will move dev work back to dedicated shops and teams when individuals and non-devs won't want to pay the AI bill to finish and ship their projects.
An outside small dev shop or internal dev team can pay these prices and spread the cost over several customers or departments, but the era of giving everyone AI and telling them to dev stuff is about to be over.
kind of sobering to realize that whether your job can be profitably automated away comes down to what $/token some hyperscale AI provider can deliver… I suppose it’s nice that this article highlights some upward pressure on that number.
Guys, we are the in the mainframe era of AI. People in the 60's thought computing was expensive too and the idea of having a computer on every desk, nevermind every pocket, nevermind every single piece of electronics in the world basically seemed like a complete pipe dream.
if you told someone in the 70's their toaster would have a supercomputer it in, they would think you were crazy. in 10 years your doorknob is going to have a local AI model it in.
This is computing 2.0 not the dot com bubble. 90% of inference will be at the edge in the future and there will still be super-computers and giant clusters doing cutting edge science and research, but for 90% of use cases youll just need a tiny local model, same reason you dont need a giant GPU in your smart tv.
The main issue with this reasoning is that the hardware substrate for AI and good old computing is the same.
All governed by Moore's Law, what happened then seems extremely unlikely to happen again, the curve is a sigmoid and we're much closer to the flat end now.
> Did we collectively forget second order thinking?
I bought 2x 16Gb NVIDIA cards this week because I don’t see hardware getting cheaper anytime soon, and because of that I totally don’t see the point of “waiting until prices go lower for graphics cards” because that might not for a long time yet!
In fact, if you include factoring in world events (and the ones that haven’t happened yet but eventually will e.g. China’s 2027 long planned take of Taiwan), then there’s no way graphics prices are going to be accessible to mere mortals until at least 2028.
But my real reasoning is that you’re going to see a flood of OpenAI and Anthropic users leave because of a) increasing pricing plans, and b) impeding business laws on the horizon about protecting sovereign data from AI (i.e data in cloud for training is a no no).
So what happens when people and companies one by one start leaving the SOTA AI cloud for from-good-enough-to-wow models? RAM and graphics cards become the new toilet paper, which is going to double again current prices.
You are incorrect. There, now we've both made unsupported assertions. Care to provide any evidence for your position?
For what it's worth, when I provide a Pangram link it's because I can already tell something is AI and I'm attempting to provide objective third-party confirmation so the conversation doesn't just degrade into me asserting that I have superior taste to you.
You are comparing two different model. It's like saying roadster is more expensive than model S. No model pricing actually increased, and I am using GPT-4o in the same price as it was before.
You can see price vs performance in artificial analysis and the the pareto optimal is all just 6 months old model.
It's hard to take this piece seriously if he's citing _Ed Zitron's_ math, and equally hard to make the blanket statement that flat-rate plans = "the current AI pricing". But yes, those pricing models were pretty silly and unsustainable.
Get back to me when there's an AI company that's actually profitable and we can compare their service and pricing.
Claiming that there's some small subset of their services (like inference per token) that's "profitable" doesn't mean anything when it relies on everything else that company is still paying for. If you could make money from it at current prices - why aren't they?
Otherwise it's just "how much they're willing to subsidize".
On OpenRouter there are 11 third-party inference providers hosting DeepSeek V4 Pro right now, of which 8 are US-based and 7 of those have zero data-retention policies (which I mention to rule out any claims of "oh they're making up the money by logging all your data"). This is a 1.6T-A49B model, so a bit bigger than Sonnet (~2/3rds the size) and a bit smaller than Opus (~3x as large). These third parties are almost perfectly interchangeable via OpenRouter as a marketplace, so they have no incentive to offer any sort of "growth pricing" below costs, and they serve it at $3.48/Mtok out.
Kimi K2.6 is 1T-A32B with a slightly less computationally efficient architecture, and is served at around $3.50/Mtok out by 9 US ZDR providers.
Unless you think that either the generally accepted size estimates for Anthropic/OpenAI models are wildly off or those companies are a lot worse at serving models efficiently, Anthropic and OpenAI are probably making around 5-8x margins on their API costs.
The cost of training new models is of course a major factor not counted here. Depending on how you want to think about that this may or may not make them net profitable. I remember one of those CEOs gave an interview a while back where they described it as a series of independent investments, where each model they train is net-positive in revenue by EOL just from its own inference, but I don't know whether that's still true or not.
Regardless, the point is that if they stopped training new models today, both Anthropic and OpenAI are making incredibly generous profits on their API inference.
The problem with fixating on earnings as you're doing is that it's a bad metric for a growth company. COGS is much more important. What you're doing is setting it up so every growth company is terrible until they've matured into a 20 year old company. That's obviously dumb.
From what I've seen pretty much every company is limited by hardware supply, to the level where's there complaints from current customers about the speed of new customer growth is exceeding their ability to service them properly.
And "growth at all costs" makes sense if there's lock in and you can monetize those "now locked-in users" later - but that doesn't really seem true on the consumer side. It seems pretty trivial to switch out which model and provider on the consumer side.
Any "lock in" has then to be on the model or inference side, and that's still advancing in multiple areas from so many different sources I'm not sure I'm comfortable saying that will also be a "winner takes all" situation either.
My approach is generally "enjoy using it while it's cheap and subsidized, but understand that might not last forever". If it does remain cheap after the subsidies end, great, you can just keep using it. But if it doesn't and you've lost the ability to work without you'll be in for a world of hurt.
Your concern about their business is that demand for their products is growing so stratospherically that they cannot meet that demand easily? I mean that's like an A+ scorecard for any business. Everyone in business would dream of such a scenario. That's called a luxury problem.
There is probably going to be a quarter or two of profits when the prices dramatically increase. Vibe coding techbros are hooked on the Iron Lung and may not want to get off.
At my work are multiple developers bragging about overnight AI usage to solve problems hands off. Yes they are wasting money and resources but the fad is here. People be vibe coding for now.
In like 6 months when all the costs need to be paid and the prices go up, we will see if these companies stay profitable. But I'm of the opinion that the vibe coding tech bros are more than enough to sustain a short or even medium term profit for these companies. Just on fad-energy alone (see OpenClaw)
The fad probably collapses soon after. I hope anyway, the waste I see is nauseating.
------------
I dunno where this is all going. But I do have faith in human ingenuity still. Things are changing, possibly for the worse, but we need to make the best of it.
The worst of behaviors is wasteful and blatant fraud. There's something useful here though.
This is where open source models are important.
The latest deepseek v4 pro model is 2-5x cheaper than Claude Sonnet 4.6. Cursor's Compose 2.5 that was just recently released is 6x cheaper than Sonnet.
The state of the art models are going to get better and more expensive and smaller models are going to get cheaper.
There will be a point where the intelligence of both the cheap and state of the art models are indistinguishable by humans like it is indistinguishable for me to understand the difference the difference between Terrance Tao and my university math professor.
I don't always need the smartest and most expensive models. I will need it every once in awhile and will gladly pay that price if I had to. What I do need is the model that will solve the current problem I have in a reasonable amount of time.
I know it comes off as pedantic to point this out but: Those are open weight models not open source models.
Closed weight models are the equivalent of SaaS. Open weight models are the equivalent of binary driver blobs or Windows software. We don't really have actual open source LLMs, which would need to publicly release their training data and technique so you could train a similar model yourself, or use their work as a baseline for your own model.
This distinction matters because an actual open source LLM would be extremely important from an ecosystem point of view, if someone ever actually released one.
There are absolutely fully open source models. These are not frontier models, but they very much do exist. OLMo is one of the models explicitly mentioned as having passed the OSI's validation phase. Pythia was also validated by the OSI as meeting its requirements for an open-source AI system. Lucie-7B is a multilingual model is one of the first LLM compliant with the OSI AI definition. Its creators explicitly state that the training dataset, data preparation code, and model weights are all publicly available under open licenses.
>The latest deepseek v4 pro model is 2-5x cheaper than Claude Sonnet 4.6. Cursor's Compose 2.5 that was just recently released is 6x cheaper than Sonnet.
It's ironic how in a thread about "AI subsidies" that people don't think free model releases from AI don't count as subsidies. Whatever AI winter that would cause AI companies to stop subsiding tokens, would probably cause other AI labs to stop doing free model releases. They might not be able to un-release the current crop of open models, but assuming proprietary model development still happens, they'll quickly go obsolete.
The currently-released models don't really go away. Even if they collectively only release a new model every few years for the sake of influence and public image, that's plenty enough to keep the competitive aspect going.
>Even if they collectively only release a new model every few years for the sake of influence and public image, that's plenty enough to keep the competitive aspect going.
This is unpersuasive. Why would AI companies (American or Chinese) stop subsidizing tokens, but keep doing open model releases? At least for the former you can argue it's a lead generation tool for enterprise contracts (eg. hobbyist uses claude code personal plan, then asks the company to buy claude code enterprise, which are billed at API rates), but what's the business case for doing open model releases? You might get some mindshare, but are also arming your competitors in the process. Moreover what makes you think the model releases will be at all competitive to frontier models? Google released gemma 4 a few weeks ago to acclaim, but it's in no way competitive to even GPT-5.4 or Opus 4.6.
> The state of the art models are going to get better and more expensive and smaller models are going to get cheaper.
Why do you think this will be true?
Right now I see the major US labs betting on gaining an advantage from having way more compute, and I see Chinese labs competing with one another in a resource-scarce environment, so they place much more emphasis on compute-efficiency.
But the supply chains that feed into the massive data center growth in the US are strained; there are energy, memory, and logistical bottlenecks to name a few.
In the medium-long run, compute capacity will not grow exponentially forever. Somehow it has for decades, but there can be no infinite exponential growth, and that point may be when the planet really starts to cook itself.
Maybe the US labs will become more compute-constrained, and then have to compete on efficiency.
Or maybe things change fundamentally in some other way I'm not thinking of.
The labs have a perverse incentive to make things as expensive compute wise as possible. The only thing keeping this somewhat in check is competition, but it's intentionally being gatekept by locking up the supply of computing infrastructure. With 3 players it's pretty easy to collude even if indirectly. They can't burn trillions forever. Nvidia's 75% profit margins are not sustainable forever.
Things will normalize, but it will take time.
>The labs have a perverse incentive to make things as expensive compute wise as possible. The only thing keeping this somewhat in check is competition, but it's intentionally being gatekept by locking up the supply of computing infrastructure. With 3 players it's pretty easy to collude even if indirectly.
By all accounts the AI capex boom is justified up by actual usage, rather than some nefarious plan for "locking up the supply of computing infrastructure". Just look at people complaining about claude availability and anthropic adding various load-shedding measures a few months ago.
>so they place much more emphasis on compute-efficiency.
Maybe on training, but on inference they use more tokens than comparable western models.
https://artificialanalysis.ai/?output-tokens=intelligence-vs...
Deepseek V4 Flash is far cheaper still, and a better model to compare to Sonnet 4.6. I'm finding it a reliable workhorse.
Yep, people who never used it say it is not good.
sorry to nitpick (I totally agree with what ur saying btw, I run Ministral-3b on my hardware as my go-to bc I don't usually need the "smartest and most expensive models")
> This is where open source models are important
open-weights, the training data isn't public
>The latest deepseek v4 pro model is 2-5x cheaper than Claude Sonnet 4.6. Cursor's Compose 2.5 that was just recently released is 6x cheaper than Sonnet.
The only way you're running Deepseek V4 with comparable quality/performance is through OpenRouter, at which point you're still susceptible to being price gouged in the future, or by spending >$20k on hardware.
> Anthropic’s CFO testified under oath this March that the company spent $10 billion on compute and made $5 billion in revenue (Ed Zitron has the math). The labs are underwater on inference. They’re raising prices to keep the lights on.
'The labs are underwater on inference' is an absurd thing to say whilst not separating the cost of _compute_ out into training and inference.
According to Dario Amodei, Anthropic are even profitable when including inference as long as you look at it on a per-model basis; it’s just that every model is more expensive to train than the last one.
For instance, if you have already spent $n to train a model and are currently earning $2n selling inference with it; but are concurrently spending $3n training the next model in anticipation of earning $6n with it, then you are already in the hole for $n and are currently also losing $n – but you are doubling your money with each model because your $n investment in the first model returns $2n and your $3n investment in the second model returns $6n.
Also:
> Ed Zitron has the math
Ed Zitron is constantly wrong about AI economics:
https://www.theargumentmag.com/p/ais-biggest-critic-has-lost...
How is training vs inference any different than other product spaces, where all the costs of bringing a product to market have to be considered for profitability? You can't just look at marginal production cost. You are still underwater if the other development costs are not being recouped by the final sales revenue.
The whole commercial AI enterprise is not economically viable if the inference revenue will not cover both inference and the amortized training costs. Given how fast they are churning through models to compete, you cannot act like the training is an asymptotically low cost.
Saying 'underwater' would have been reasonable, but 'underwater on inference' is a nonsense way to say it.
I mean, I guess they could just stop training new models and coast, but they ARE training models so you have to include those costs.
I wonder how much of Uber blowing their AI budget and MSFT pulling their claude code licenses can be attributed to "tokenmaxxing".
When Meta announced token leaderboards and other followed, I could see this being the logical conclusion. That whole trend is so dumb because it leads to this.
Company announces they will measure developer performance by how many tokens they burn and constantly talks about how the best developers burn the most tokens. Developers see the message and start burning tokens. And then the company acts surprised when their bills go through the roof.
I personally use my OpenAI subscription pretty heavily, 2-3 agents running practically all day on various tasks but I never even get close to running into limits while I hear about others blowing through limits on multiple accounts in the same time period. I'm convinced that most of those folks and their elaborate workflows aren't really for productivity but for bragging rights about how much they use AI.
The same here, where I haven't come close to hitting any of my CC limits. Even though I'm more productive than I've ever been (as measured by finished, valuable tasks running in production) and I'm clearing out months of backlog, I have either one of two conclusions when I hear about others who suggest they need more:
1. I'm doing it wrong. Apparently I'm supposed to give it a vague paragraph about what the business does, and I can run off and sip margaritas and wake up to a fully fleshed business
2. They don't know what they're doing, and they're sending the LLM off on a wild goose chase that it does a reasonable job of working it's way out of, so they consider it success despite the waste.
> I personally use my OpenAI subscription pretty heavily, 2-3 agents running practically all day on various tasks but I never even get close to running into limits
Same. But if I was working for an organization that measured token usage, you can bet I would be doing things like creating a cron job that uses claude to create a customized bespoke report update of the current status of all my open assigned tickets and message that to myself 4 times a day... token burn for zero purpose whatsoever.
> I'm convinced that most of those folks and their elaborate workflows aren't really for productivity but for bragging rights about how much they use AI.
This is quite the reductive, charged statement. Can I ask what subscription plan you're using?
My personal experience is unlike this at all-- I work on ever-expanding codebases so I can easily burn tokens. Not to mention, structured agentic coding with adverserial reviews & task organization is not token-efficient. Additionally, for the problems I'm working on, only xhigh or high reasoning gives me worthwhile results while saving time. There are definitely configurations where default consumption doesn't work.
For reference, I used 15 billion tokens (most of it cached) last month on my day job's enterprise plan. That doesn't include my personal plans' usage.
I'm on OpenAI's Pro (200/mo) plan.
I make like 2 prompts a week to gemini flash on the weband get more done than all the people that are exhibiting literal manic behavior in the way they use LLMs.
I really wish the management behind these dumb ideas couldnt just quietly pretend they never did it once it goes out of fashion.
The fact that somebody established a leaderboard for tokenmaxxing ought to follow you around like a black cloud for the rest of your career once the collective hallucination lifts and people realize just how monumentally stupid it was.
Alas they do all these stupid things together which makes it seem more defensible and then everybody forgets.
In the three options OP presents, I wonder if there's a fourth: BYO model
Customers give vendors metered access to their model. They can budget tokens per vendor. Vendors selling "AI products" can have a cleaner story and win on the margin.
The first step to is to iron out a reasonable protocol, basically authorizing a, access token, and then the model providers (OpenAI, Anthropic, etc.) do the rate limiting. Theoretically this could be done by OpenRouter too.
But even so - do customers want an "AI product" packaged cleanly, or do they want to manage token capacity? They may be forced to do the latter....
The math seems off. How is 7.8 million vs 4 million 95% more expensive. Article makes good points but I doubt the numbers as they don’t add up.
Still agree with the conclusion though.
What is the OP talking about. $/unit intelligence is going down rapidly. You can achieve what would have been considered miracles in 2022 with < $10.
Absolutely, though I think the expectations are being set by those who have watched too many "OpenClaw business on autopilot" videos.
Not mentioned in the article/blog was the local alternative. Many applications will run just fine locally and not in the cloud. This is also more secure. Running local will probably eventually become the norm. It makes me wonder about the future of all these VC funded AI companies...
I seldom use my PC anymore ever since i got a laptop. with the cost per token increasing along with the random "features" where models will just eat through your tokens in one hour. I really have been tempted to turn my PC into a server to run local models on there
Insofar as I can tell, inference is on a certain path toward becoming "free". The models are now extremely powerful on high-end consumer hardware, and the efficiency trend seems likely to continue.
Here is a recent non-rigorous benchmark I ran against a bunch of models. Qwen3.6 35B A3B fine-tuned with opus data runs plenty fast on my local machine and produce outstanding results - easily in the top 5, comparable to GPT 5.5 Pro (which is $180/mtok).
https://gistpreview.github.io/?31d66ef69e4aed3efae1aec69d86c...
I've predicted for years now that the industry will head down the path of the virus scanning vendors: selling subscriptions to be able to download the latest versions of models. I simply don't see how any other business model is remotely viable, except at the very highest end of inference or video gen.
That local hardware is not consumer though but prosumer. Consumer is a 500$ laptop running that and that is not currently the case.
Has this not been true for a long time now? Most companies have had enterprise/business level prices that was highly connected to usage for a what feels like at least a year.
This is only true if your world is limited to openai, antropic and alike.
There are a whole bunch of companies somewhere else in the world that are getting better and cheaper every month, hardware side included. all without the infinite VC money
My expectation is that local models will be the default for coding within a year or two. You can already run Qwen 3.6 with MTP at a pretty reasonable speed without needing a huge amount of VRAM. And while it's not as good as current frontier models, it's already quite competent for a lot of tasks.
And there's no sign that people are running out of ideas for how to optimize models further. You see a bunch of papers come out literally every few weeks right now. So, it's entirely plausible to me that we'll see models that are superior to current frontier ones in a year or two that will run on your machine.
Once we get to that point, I don't think it's even going to matter if frontier models keep improving for most people. Being able to run the model on your machine, use it as much as you want in any way you want, without having to worry about it changing from under you or the company changing pricing, and not have to send all your data to the vendor are going to be the deciding factors.
At some point the models are just good enough to do what you need to do. On top of that, I expect tooling around models and coding patterns will evolve as well. That could compensate significantly for the capabilities of the model. We already see this happening with two prime examples here:
https://github.com/itigges22/ATLAS
https://arxiv.org/abs/2509.16198
Capex and revenue should not be compared like this, unless revenue is small and not growing.
I get similar results for deepseek and opus but opus is way faster. I guess deepseek streams thinking and makes it slower?
Inference costs absolutely did fall. And even more so when looking at intelligence it buys you.
eg compare say gpt 3.5 to latest deepseek. Both cheaper and more at more capable
Some of these coming price increases will move dev work back to dedicated shops and teams when individuals and non-devs won't want to pay the AI bill to finish and ship their projects.
An outside small dev shop or internal dev team can pay these prices and spread the cost over several customers or departments, but the era of giving everyone AI and telling them to dev stuff is about to be over.
kind of sobering to realize that whether your job can be profitably automated away comes down to what $/token some hyperscale AI provider can deliver… I suppose it’s nice that this article highlights some upward pressure on that number.
Guys, we are the in the mainframe era of AI. People in the 60's thought computing was expensive too and the idea of having a computer on every desk, nevermind every pocket, nevermind every single piece of electronics in the world basically seemed like a complete pipe dream.
if you told someone in the 70's their toaster would have a supercomputer it in, they would think you were crazy. in 10 years your doorknob is going to have a local AI model it in.
This is computing 2.0 not the dot com bubble. 90% of inference will be at the edge in the future and there will still be super-computers and giant clusters doing cutting edge science and research, but for 90% of use cases youll just need a tiny local model, same reason you dont need a giant GPU in your smart tv.
The main issue with this reasoning is that the hardware substrate for AI and good old computing is the same.
All governed by Moore's Law, what happened then seems extremely unlikely to happen again, the curve is a sigmoid and we're much closer to the flat end now.
> Memory for 4x expensive
> Did we collectively forget second order thinking?
I bought 2x 16Gb NVIDIA cards this week because I don’t see hardware getting cheaper anytime soon, and because of that I totally don’t see the point of “waiting until prices go lower for graphics cards” because that might not for a long time yet!
In fact, if you include factoring in world events (and the ones that haven’t happened yet but eventually will e.g. China’s 2027 long planned take of Taiwan), then there’s no way graphics prices are going to be accessible to mere mortals until at least 2028.
But my real reasoning is that you’re going to see a flood of OpenAI and Anthropic users leave because of a) increasing pricing plans, and b) impeding business laws on the horizon about protecting sovereign data from AI (i.e data in cloud for training is a no no).
So what happens when people and companies one by one start leaving the SOTA AI cloud for from-good-enough-to-wow models? RAM and graphics cards become the new toilet paper, which is going to double again current prices.
Upgrade now before it’s too late folks!
This is slightly more tasteful slop than average (I'm thinking probably Claude rather than ChatGPT?), but it's still 100% AI written: https://www.pangram.com/history/c55ab69b-e0a9-49a0-8056-2fcd...
This... is not a reliable AI detection method at all.
You are incorrect. There, now we've both made unsupported assertions. Care to provide any evidence for your position?
For what it's worth, when I provide a Pangram link it's because I can already tell something is AI and I'm attempting to provide objective third-party confirmation so the conversation doesn't just degrade into me asserting that I have superior taste to you.
Pangram is highly reliable.
You are comparing two different model. It's like saying roadster is more expensive than model S. No model pricing actually increased, and I am using GPT-4o in the same price as it was before.
You can see price vs performance in artificial analysis and the the pareto optimal is all just 6 months old model.
It's hard to take this piece seriously if he's citing _Ed Zitron's_ math, and equally hard to make the blanket statement that flat-rate plans = "the current AI pricing". But yes, those pricing models were pretty silly and unsustainable.
Get back to me when there's an AI company that's actually profitable and we can compare their service and pricing.
Claiming that there's some small subset of their services (like inference per token) that's "profitable" doesn't mean anything when it relies on everything else that company is still paying for. If you could make money from it at current prices - why aren't they?
Otherwise it's just "how much they're willing to subsidize".
On OpenRouter there are 11 third-party inference providers hosting DeepSeek V4 Pro right now, of which 8 are US-based and 7 of those have zero data-retention policies (which I mention to rule out any claims of "oh they're making up the money by logging all your data"). This is a 1.6T-A49B model, so a bit bigger than Sonnet (~2/3rds the size) and a bit smaller than Opus (~3x as large). These third parties are almost perfectly interchangeable via OpenRouter as a marketplace, so they have no incentive to offer any sort of "growth pricing" below costs, and they serve it at $3.48/Mtok out.
Kimi K2.6 is 1T-A32B with a slightly less computationally efficient architecture, and is served at around $3.50/Mtok out by 9 US ZDR providers.
Unless you think that either the generally accepted size estimates for Anthropic/OpenAI models are wildly off or those companies are a lot worse at serving models efficiently, Anthropic and OpenAI are probably making around 5-8x margins on their API costs.
The cost of training new models is of course a major factor not counted here. Depending on how you want to think about that this may or may not make them net profitable. I remember one of those CEOs gave an interview a while back where they described it as a series of independent investments, where each model they train is net-positive in revenue by EOL just from its own inference, but I don't know whether that's still true or not.
Regardless, the point is that if they stopped training new models today, both Anthropic and OpenAI are making incredibly generous profits on their API inference.
The problem with fixating on earnings as you're doing is that it's a bad metric for a growth company. COGS is much more important. What you're doing is setting it up so every growth company is terrible until they've matured into a 20 year old company. That's obviously dumb.
From what I've seen pretty much every company is limited by hardware supply, to the level where's there complaints from current customers about the speed of new customer growth is exceeding their ability to service them properly.
And "growth at all costs" makes sense if there's lock in and you can monetize those "now locked-in users" later - but that doesn't really seem true on the consumer side. It seems pretty trivial to switch out which model and provider on the consumer side.
Any "lock in" has then to be on the model or inference side, and that's still advancing in multiple areas from so many different sources I'm not sure I'm comfortable saying that will also be a "winner takes all" situation either.
My approach is generally "enjoy using it while it's cheap and subsidized, but understand that might not last forever". If it does remain cheap after the subsidies end, great, you can just keep using it. But if it doesn't and you've lost the ability to work without you'll be in for a world of hurt.
Your concern about their business is that demand for their products is growing so stratospherically that they cannot meet that demand easily? I mean that's like an A+ scorecard for any business. Everyone in business would dream of such a scenario. That's called a luxury problem.
There is probably going to be a quarter or two of profits when the prices dramatically increase. Vibe coding techbros are hooked on the Iron Lung and may not want to get off.
At my work are multiple developers bragging about overnight AI usage to solve problems hands off. Yes they are wasting money and resources but the fad is here. People be vibe coding for now.
In like 6 months when all the costs need to be paid and the prices go up, we will see if these companies stay profitable. But I'm of the opinion that the vibe coding tech bros are more than enough to sustain a short or even medium term profit for these companies. Just on fad-energy alone (see OpenClaw)
The fad probably collapses soon after. I hope anyway, the waste I see is nauseating.
------------
I dunno where this is all going. But I do have faith in human ingenuity still. Things are changing, possibly for the worse, but we need to make the best of it.
The worst of behaviors is wasteful and blatant fraud. There's something useful here though.