Claude subscription is equivalant of spot instance
And APIs are on-demand service equivalant.
Priority is set to APIs and leftover compute is used by Subscription Plans.
When there is no capacity, subscriptions are routed to Highly Quantized cheaper models behind the scenes.
Selling subscription makes it cheaper to run such inference otherwise many times your capacity is just sitting there idle.
I've used qwen code, codex and claude.
Codex is 2x better than Qwen code and Claude is 2x better than Codex.
So I'd hope the Claude Opus is atleast 4-5x more expensive to run than flagship Qwen Code model hosted by Alibaba.
overrun11 6 minutes ago [-]
A huge number of people are convinced that OpenAI and Anthropic are selling inference tokens at a loss despite the fact that there's no evidence this is true and a lot of evidence that it isn't. It's just become a meme uncritically regurgitated.
This sloppy Forbes article has polluted the epistemic environment because now theres a source to point to as "evidence."
So yes this post author's estimation isn't perfect but it is far more rigorous than the original Forbes article which doesn't appear to even understand the difference between Anthropic's API costs and its compute costs.
hirako2000 2 hours ago [-]
> Qwen 3.5 397B-A17B is a good comparison
It is not. It's a terrible comparison. Qwen, deepseek and other Chinese models are known for their 10x or even better efficiency compared to Anthropic's.
That's why the difference between open router prices and those official providers isn't that different. Plus who knows what open routed providers do in term quantization. They may be getting 100x better efficiency, thus the competitive price.
That being said not all users max out their plan, so it's not like each user costs anthropic 5,000 USD. The hemoragy would be so brutal they would be out of business in months
jychang 2 hours ago [-]
That's a tautology. People think chinese models are 10x more efficient because they're 10x cheaper, and then you use that to claim that they're 10x more efficient.
Opus isn't that expensive to host. Look at Amazon Bedrock's t/s numbers for Opus 4.5 vs other chinese models. They're around the same order of magnitude- which means that Opus has roughly the same amount of active params as the chinese models.
Also, you can select BF16 or Q8 providers on openrouter.
re-thc 20 minutes ago [-]
> That's a tautology. People think chinese models are 10x more efficient because they're 10x cheaper
They do have different infrastructure / electricity costs and they might not run on nvidia hardware.
It's not just the models.
jychang 10 minutes ago [-]
Except there are providers that serve both chinese models AND opus as well. On the same hardware.
Namely, Amazon Bedrock and Google Vertex.
That means normalized infrastructure costs, normalized electricity costs, and normalized hardware performance. Normalized inference software stack, even (most likely). It's about a close of a 1 to 1 comparison as you can get.
Both Amazon and Google serve Opus at roughly ~1/2 the speed of the chinese models. Note that they are not incentivized to slow down the serving of Opus or the chinese models! So that tells you the ratio of active params for Opus and for the chinese models.
Weaver_zhu 10 minutes ago [-]
Agree, but I guess the Opus 4.6 is 10x larger, rather than Chinese models being 10x more efficient. It is said that GPT-4 is already a 1.6T model, and Llama 4 behemoth is also much bigger than Chinese open-weight models. Chinese tech companies are short of frontier GPUs, but they did a lot of innovations on inference efficiency (Deepseek CEO Liang himself shows up in the author list of the related published papers).
jychang 5 minutes ago [-]
No, Opus cannot be 10x larger than the chinese models.
If Opus was 10x larger than the chinese models, then Google Vertex/Amazon Bedrock would serve it 10x slower than Deepseek/Kimi/etc.
That's not the case. They're in the same order of magnitude of speed.
lelanthran 2 hours ago [-]
> That being said not all users max out their plan,
These are not cell phone plans which the average joe takes, they are plans purchased with the explicit goal of software development.
I would guess that 99 out of every 100 plans are purchased with the explicit goal of maxing them out.
serial_dev 2 hours ago [-]
I’m not maxing them out… I have issues that I need to fix, features I need to develop, and I have things I want to learn.
When I have a feeling that these tools will speed me up, I use them.
My client pays for a couple of these tools in an enterprise deal, and I suspect most of us on the team work like that.
If my goal was to max out every tool my client pays, I’d be working 24hrs a day and see no sunlight ever.
I guess it’s like the all you can eat buffet. Everybody eats a lot, but if you eat so much that you throw up and get sick, you are special.
bloppe 15 minutes ago [-]
I am special
Ginden 1 hours ago [-]
My employer bought me a Claude Max subscription. On heavy weeks I use 80% of the subscription. And among software engineers that I know, I'm a relatively heavy user.
Why? Because in my experience, the bottleneck is in shareholders approving new features, not my ability to dish out code.
raihansaputra 1 hours ago [-]
goal? yeah. but in reality just timing it right (starting a session at 7-8am, to get 2 sessions in a workday, or even 3 if you can schedule something at 5am), i rarely hit limits.
if i hit the limit usually i'm not using it well and hunting around. if i'm using it right i'm basically gassed out trying to hit the limit to the max.
solumunus 1 hours ago [-]
There’s absolutely no way that’s true.
rustystump 1 hours ago [-]
In saas this is not true. Most saas is highly profitable or was i suppose because they knew that most of their customers would never max out their plans.
simianwords 2 hours ago [-]
>It is not. It's a terrible comparison. Qwen, deepseek and other Chinese models are known for their 10x or even better efficiency compared to Anthropic's.
I find it a good comparison because it is a good baseline since we have zero insider knowledge of Anthropic. They give me an idea that a certain size of a model has a certain cost associated.
I don't buy the 10x efficiency thing: they are just lagging behind the performance of current SOTA models. They perform much worse than the current models while also costing much less - exactly what I would expect. Current Qwen models perform as good as Sonnet 3 I think. 2 years later when Chinese models catchup with enough distillation attacks, they would be as good as Sonnet 4.6 and still be profitable.
ineedaj0b 4 minutes ago [-]
What CC costs internally is not public. How efficient it is, is not public.
…You could take efficiency improvement rates from previous models releases (from x -> y) and assume; they have already made “improvements” internally. This is likely closer to what their real costs are.
anonzzzies 1 hours ago [-]
I calculated only last weekend that my team would cost, if we would run Claude Code on retail API costs, around $200k/mo. We pay $1400/month in Max subscriptions. So that's $50k/user... But what tokens CC is reporting in their json -> a lot of this must be cached etc, so doubt it's anywhere near $50k cost, but not sure how to figure out what it would cost and I'm sure as hell not going to try.
scandox 36 minutes ago [-]
I'm fascinated to know the kind of work that allows you to intelligently allocate so much resources. I use Claude extensively and feel that I great value out of it but I reach a limit in terms of what I can do that makes sense relatively quickly it seems.
aweb 9 minutes ago [-]
I'm surprised, isn't it forbidden to use the Max plan as part of a company? Just curious, as I thought it was forbidden by the ToS but I'm not sure if I have a good understanding of it
quikoa 2 minutes ago [-]
If they believe a sufficient number is locked in then they may consider doing this later.
bloppe 8 minutes ago [-]
If that were true, then everyone I know is violating that tos
neamar 1 hours ago [-]
You can use `npx ccusage` to check your local logs and see how much it would have cost through the API.
jychang 1 hours ago [-]
> but not sure how to figure out what it would cost and I'm sure as hell not going to try.
Ask Opus to figure out how much it would cost. Lol.
eaglelamp 3 hours ago [-]
If Anthropic's compute is fully saturated then the Claude code power users do represent an opportunity cost to Anthropic much closer to $5,000 then $500.
Anthropic's models may be similar in parameter size to model's on open router, but none of the others are in the headlines nearly as much (especially recently) so the comparison is extremely flawed.
The argument in this article is like comparing the cost of a Rolex to a random brand of mechanical watch based on gear count.
d1sxeyes 2 hours ago [-]
But opportunity cost is not actual cost. “If everyone just kept paying but used our service less we would be more profitable” is true, but not in any meaningful way.
Are Anthropic currently unable to sell subscriptions because they don’t have capacity?
MaxikCZ 51 minutes ago [-]
> Are Anthropic currently unable to sell subscriptions because they don’t have capacity?
Absolutely! Im currently paying $170 to google to use Opus in antigravity without limit in full agent mode, because I tried Anthropic $20 subscription and busted my limit within a single prompt. Im not gonna pay them $200 only to find out I hit the limit after 20 or even 50 prompts.
And after 2 more months my price is going to double to over $300, and I still have no intention of even trying the 20x Max plan, if its really just 20x more prompts than Pro.
dtech 32 minutes ago [-]
This has a absolutely nothing to do with whether they're limited by available compute...
MaxikCZ 17 minutes ago [-]
What? Wouldn't they give me more than 1 prompt of compute for my $20, if they had spare?
esrauch 5 minutes ago [-]
I don't think that logically follows.
They have a business model and are trying to capture more revenue, fully saturating your computer isn't obviously a good business strategy.
eru 54 minutes ago [-]
Opportunity costs are real. In many cases they are more real than 'actual costs'. However, I otherwise agree with you.
Aeolun 2 hours ago [-]
Opportunity cost is not the same thing as actual cost. They might have made more money if they were capable of selling the API instead of CC, but I would never tell my company to use CC all the time if I didn’t have a personal subscription.
eaglelamp 2 hours ago [-]
You’re looking through the wrong end of the telescope. An investor is buying opportunity and it is a real cost to them.
kaliqt 36 minutes ago [-]
Still makes no sense as they’d lose revenue, data, and scale if they don’t subsidize.
nottorp 21 minutes ago [-]
You know who also loves to use the term "opportunity cost"?
The entertainment industry. They still tell you about how much money they're leaving on the table because people pirate stuff.
What would happen in reality for entertainment is people would "consume" far less "content".
And what would happen in reality for Anthropic is people would start asking themselves if the unpredictability is worth the price. Or at best switch to pay as you go and use the API far less.
bob1029 2 hours ago [-]
> If Anthropic's compute is fully saturated then the Claude code power users do represent an opportunity cost to Anthropic much closer to $5,000 then $500.
I think it's the other way around? Sparse use of GPU farms should be the more expensive thing. Full saturation means that we can exploit batching effects throughout.
KronisLV 2 hours ago [-]
Don’t give them any ideas, please! I need my 100 USD subscription with generous Opus usage!
eru 54 minutes ago [-]
Google's Antigravity has Opus access, and I suspect it's subsidised.
NooneAtAll3 2 hours ago [-]
> The argument in this article is like comparing the cost of a Rolex to a random brand of mechanical watch on gear count
I mean... rolex is overpriced brand whose cost to consumers is mainly just marketting in itself. Its production cost is nowhere close to selling price and looking at gears is fair way of evaluating that
fragmede 1 hours ago [-]
> production cost is nowhere close to selling price
When has production cost had anything to do with selling price?
eru 53 minutes ago [-]
Not directly. But if production cost is above selling price, you typically tend to get less production. And if production cost is (way) below selling price, that tends to invite competition.
YetAnotherNick 2 hours ago [-]
You can rent the GPUs and everything needed to run the model. Opportunity cost is not a real cost here.
Only thing that matters is if the users would have paid $5000 if they don't have option to buy subscription. And I highly doubt they would have.
ymaws 4 hours ago [-]
How confident are you in the opus 4.6 model size? I've always assumed it was a beefier model with more active params that Qwen397B (17B active on the forward pass)
Bolwin 3 hours ago [-]
Yeah that's a massive assumption they're making. I remember musk revealed Grok was multiple trillion parameters. I find it likely Opus is larger.
I'm sure Anthropic is making money off the API but I highly doubt it's 90% profit margins.
jychang 2 hours ago [-]
> I find it likely Opus is larger.
Unlikely. Amazon Bedrock serves Opus at 120tokens/sec.
If you want to estimate "the actual price to serve Opus", a good rough estimate is to find the price max(Deepseek, Qwen, Kimi, GLM) and multiply it by 2-3. That would be a pretty close guess to actual inference cost for Opus.
It's impossible for Opus to be something like 10x the active params as the chinese models. My guess is something around 50-100b active params, 800-1600b total params. I can be off by a factor of ~2, but I know I am not off by a factor of 10.
simianwords 2 hours ago [-]
Are you sure you can use tps as a proxy?
jychang 1 hours ago [-]
In practice, tps is a reflection of vram memory bandwidth during inference. So the tps tells you a lot about the hardware you're running on.
Comparing tps ratios- by saying a model is roughly 2x faster or slower than another model- can tell you a lot about the active param count.
I won't say it'll tell you everything; I have no clue what optimizations Opus may have, which can range from native FP4 experts to spec decoding with MTP to whatever. But considering chinese models like Deepseek and GLM have MTP layers (no clue if Qwen 3.5 has MTP, I haven't checked since its release), and Kimi is native int4, I'm pretty confident that there is not a 10x difference between Opus and the chinese models. I would say there's roughly a 2x-3x difference between Opus 4.5/4.6 and the chinese models at most.
fc417fc802 28 minutes ago [-]
> In practice, tps is a reflection of vram memory bandwidth during inference.
> Comparing tps ratios- by saying a model is roughly 2x faster or slower than another model- can tell you a lot about the active param count.
You sure about that? I thought you could shard between GPUs along layer boundaries during inference (but not training obviously). You just end up with an increasingly deep pipeline. So time to first token increases but aggregate tps also increases as you add additional hardware.
jychang 16 minutes ago [-]
That doesn't work. Think about it a bit more.
Hint: what's in the kv cache for the 2nd token?
And that's called layer parallelism (as opposed to tensor parallelism). It allows you to run larger models (pooling vram across gpus) but does not allow you to run models faster.
Tensor parallelism DOES allow you to run models faster across multiple GPUs, but you're limited to how fast you can synchronize the all-reduce. And in general, models would have the same boost on the same hardware- so the chinese models would have the same perf multiplier as Opus.
nbardy 1 hours ago [-]
You can estimate on tok/second
The Trillions of parameters claim is about the pretraining.
It’s most efficient in pre training to train the biggest models possible. You get sample efficiency increase for each parameter increase.
However those models end up very sparse and incredibly distillable.
And it’s way too expensive and slow to serve models that size so they are distilled down a lot.
2 hours ago [-]
aurareturn 2 hours ago [-]
Anthropic CEO said 50%+ margins in an interview. I'm guessing 50 - 60% right now.
daemonologist 4 hours ago [-]
Even if it's larger, OpenRouter has DeepSeek v3.2 (685B/37B active) at $0.26/0.40 and Kimi K2.5 (1T/32B active) at $0.45/2.25 (mentioned in the post).
johndough 3 hours ago [-]
Opus 4.6 likely has in the order of 100B active parameters. OpenRouter lists the following throughput for Google Vertex:
42 tps for Claude Opus 4.6 https://openrouter.ai/anthropic/claude-opus-4.6
143 tps for GLM 4.7 (32B active parameters) https://openrouter.ai/z-ai/glm-4.7
70 tps for Llama 3.3 70B (dense model) https://openrouter.ai/meta-llama/llama-3.3-70b-instruct
For GLM 4.7, that makes 143 * 32B = 4576B parameters per second, and for Llama 3.3, we get 70 * 70B = 4900B, which makes sense since denser models are easier to optimize. As a lower bound, we get 4576B / 42 ≈ 109B active parameters for Opus 4.6. (This makes the assumption that all three models use the same number of bits per parameter and run on the same hardware.)
jychang 2 hours ago [-]
Yep, you can also get similar analysis from Amazon Bedrock, which serves Opus as well.
I'd say Opus is roughly 2x to 3x the price of the top Chinese models to serve, in reality.
codemog 4 hours ago [-]
Also curious if any experts can weigh in on this. I would guess in the 1 trillion to 2 trillion range.
Chamix 2 hours ago [-]
Try 10s of trillions. These days everyone is running 4-bit at inference (the flagship feature of Blackwell+), with the big flagship models running on recently installed Nvidia 72gpu rubin clusters (and equivalent-ish world size for those rented Ironwood TPUs Anthropic also uses). Let's see, Vera Rubin racks come standard with 20 TB (Blackwell NVL72 with 10 TB) of unified memory, and NVFP4 fits 2 parameters per btye...
Of course, intense sparsification via MoE (and other techniques ;) ) lets total model size largely decouple from inference speed and cost (within the limit of world size via NVlink/TPU torrus caps)
So the real mystery, as always, is the actual parameter count of the activated head(s). You can do various speed benchmarks and TPS tracking across likely hardware fleets, and while an exact number is hard to compute, let me tell you, it is not 17B or anywhere in that particular OOM :)
Comparing Opus 4.6 or GPT 5.4 thinking or Gemini 3.1 pro to any sort Chinese model (on cost) is just totally disingenuous when China does NOT have Vera Rubin NVL72 GPUs or Ironwood V7 TPUs in any meaningful capacity, and is forced to target 8gpu Blackwell systems (and worse!) for deployment.
jychang 1 hours ago [-]
Nobody is running 10s of trillion param models in 2026. That's ridiculous.
Opus is 2T-3T in size at most.
aurareturn 2 hours ago [-]
China is targeting H20 because that's all they were officially allowed to buy.
Chamix 2 hours ago [-]
I generally agree, back of the napkin math shows H20 cluster of 8gpu * 96gb = 768gb = 768B parameters on FP8 (no NVFP4 on Hopper), which lines up pretty nicely with the sizes of recent open source Chinese models.
However, I'd say its relatively well assumed in realpolitik land that Chinese labs managed to acquire plenty of H100/200 clusters and even meaningful numbers of B200 systems semi-illicitly before the regulations and anti-smuggling measures really started to crack down.
This does somewhat beg the question of how nicely the closed source variants, of undisclosed parameter counts, fit within the 1.1tb of H200 or 1.5tb of B200 systems.
aurareturn 1 hours ago [-]
They do not have enough H200 or Blackwell systems to server 1.6 billion people and the world so I doubt it's in any meaningful number.
Chamix 1 hours ago [-]
I assure you, the number of people paying to use Qwen3-Max or other similar proprietary endpoints is far less than 1.6 billion.
aurareturn 1 hours ago [-]
You don't need to assure me. It's a theoretical maximum.
himata4113 58 minutes ago [-]
What people don't realize is that cache is *free*, well not free, but compared to the compute required to recompute it? Relatively free.
If you remove the cached token cost from pricing the overall api usage drops from around $5000 to $800 (or $200 per week) on the $200 max subscription. Still 4x cheaper over API, but not costing money either - if I had to guess it's break even as the compute is most likely going idle otherwise.
eru 55 minutes ago [-]
> [...] if I had to guess it's break even as the compute is most likely going idle otherwise.
Why would it go idle? It would go to their next best use. At least they could help with model training or let their researchers run experiments etc.
himata4113 53 minutes ago [-]
inference compute is vastly different versus training, also it has to stay hot in vram which probably takes up most of it. There is limited use for THAT much compute as well, they are running things like claude code compiler and even then they're scratching the surface of the amount of compute they have.
Training currently requires nvidia's latest and greatest for the best models (they also use google TPU's now which are also technically the latest and greatest? However, they're more of a dual purpose than anything afaik so that would be a correct assesment in that case)
Inference can run on a hot potato if you really put your mind to it
eru 52 minutes ago [-]
They can run any number of inference experiments. Like a lot of the alignment work they have going on.
I am not saying this would be a great use of their compute, but idle is far from the only alternative. (Unless electricity is the binding constraint?)
himata4113 50 minutes ago [-]
Electricity is charged whenever you use it or not, so very unlikely, but sure, they can find uses for it. Although they are not going to make that much money compared to claude code subscriptions.
jeff_antseed 2 hours ago [-]
the openrouter comparison is interesting because it shows what happens when you have actual supply-side competition. multiple providers, different quantizations, price competition. the spread between cheapest and priciest for the same model can be 3-5x.
anthropic doesn't have that. single provider, single pricing decision. whether or not $5k is accurate the more interesting question is what happens to inference pricing when the supply side is genuinely open. we're seeing hints of it with open router but its still intermediated
not saying this solves anthropic's cost problem, just that the "what does inference actually cost" question gets a lot more interesting when providers are competing directly
z3ugma 5 hours ago [-]
This is such a well-written essay. Every line revealed the answer to the immediate question I had just thought of
lovecg 4 hours ago [-]
I can’t get past all the LLM-isms. Do people really not care about AI-slopifying their writing? It’s like learning about bad kerning, you see it everywhere.
crakhamster01 1 hours ago [-]
I had a similar reaction to OP for a different post a few weeks back - I think some analysis on the health economy. Initially as I was reading I thought - "Wow, I've never read a financial article written so clearly". Everything in layman's terms. But as I continued to read, I began to notice the LLM-isms. Oversimplified concepts, "the honest truth" "like X for Y", etc.
Maybe the common factor here is not having deep/sufficient knowledge on the topic being discussed? For the article I mentioned, I feel like I was less focused on the strength of the writing and more on just understanding the content.
LLMs are very capable at simplifying concepts and meeting the reader at their level. Personally, I subscribe to the philosophy of - "if you couldn't be bothered to write it, I shouldn't bother to read it".
ajkjk 22 minutes ago [-]
Alternate theory... a few months into the LLMism phenomenon, people are starting to copy the LLM writing style without realizing it :(
weird-eye-issue 3 hours ago [-]
I think you're just hallucinating because this does not come across as an AI article
lovecg 3 hours ago [-]
I see quite a few:
“what X actually is”
“the X reality check”
Overuse of “real” and “genuine”:
> The real story is actually in the article. … And the real issue for Cursor … They have real "brand awareness", and they are genuinely better than the cheaper open weights models - for now at least. It's a real conundrum for them.
> … - these are genuinely massive expenses that dwarf inference costs.
This style just screams “Claude” to me.
hansvm 3 hours ago [-]
It was almost certainly at least heavily edited with one. Ignoring the content, every single thing about the structure and style screams LLM.
lelanthran 2 hours ago [-]
> I think you're just hallucinating because this does not come across as an AI article
It has enough tells in the correct frequency for me to consider it more than 50% generated.
NetOpWibby 3 hours ago [-]
Name checks out
Erem 3 hours ago [-]
I don’t see the usual tells in this essay
rhubarbtree 2 hours ago [-]
It is certainly very obvious a lot of the time. I wonder if we revisited the automated slop detection problem we’d be more successful now… it feels like there are a lot more tells and models have become more idiosyncratic.
weird-eye-issue 1 hours ago [-]
Tons of companies do this already. It's not like this is a problem that nobody is constantly revisiting...
152334H 3 hours ago [-]
People care, when they can tell.
Popular content is popular because it is above the threshold for average detection.
In a better world, platforms would empower defenders, by granting skilled human noticers flagging priority, and by adopting basic classifiers like Pangram.
Unfortunately, mainstream platforms have thus far not demonstrated strong interest in banning AI slop. This site in particular has actually taken moderation actions to unflag AI slop, in certain occasions...
n_u 3 hours ago [-]
Good article! Small suggestions:
1. It would be nice to define terms like RSI or at least link to a definition.
2. I found the graph difficult to read. It's a computer font that is made to look hand-drawn and it's a bit low resolution. With some googling I'm guessing the words in parentheses are the clouds the model is running on. You could make that a bit more clear.
brianjeong 4 hours ago [-]
These margins are far greater than the ones Dario has indicated during many of his recent podcasts appearances.
skybrian 3 hours ago [-]
What did he say?
aurareturn 3 hours ago [-]
By the way, one of the charts in the article shows that Opus 4.6 is 10x costlier than Kimi K2.5.
I thought there was no moat in AI? Even being 10x costlier, Anthropic still doesn't have enough compute to meet demand.
Those "AI has no moat" opinions are going to be so wrong so soon.
spiderice 3 hours ago [-]
Claude Code Max obviously doesn't cost 10x more than Kimi. The article even confirms that you can get $5k worth of computer for $200 with Claude Code Max.
So no, Claude would not be getting NEARLY as much usage as it's currently getting if it weren't for the $100/$200 monthly subscription. You're comparing Kimi to the price that most people aren't paying.
hattmall 3 hours ago [-]
Is it fair to say the Open Router models aren't subsidized though? They make the case that companies on there are running a business, but there are free models, and companies with huge AI budgets that want to gather training data and show usage.
functionmouse 9 hours ago [-]
Was anyone under the impression that it does? Serious question. I've never heard that, personally.
versteegen 3 hours ago [-]
Ed Zitron made that claim (in particular here: [1]). In the same article he admits he not a programmer, and had to ask someone else to try out Claude Code and ccusage for him. He doesn't have any understanding of how LLMs or caching works. But he's prominent because he's received leaked financial details for Anthropic and OpenAI, eg [2]
You would be surprised because there are lots of posters here who think that the cost is so enormous that this whole industry is unviable.
crazygringo 4 hours ago [-]
I mean, the very first paragraph of TFA is describing who is under that impression. Literally the first sentence:
> My LinkedIn and Twitter feeds are full of screenshots from the recent Forbes article on Cursor claiming that Anthropic's $200/month Claude Code Max plan can consume $5,000 in compute.
fulafel 3 hours ago [-]
That's claiming that worst case, a subscriber _can_ use that much. It's possible that's wrong too, but in any case a lot of services are built on the assumption that the average user doesn't max out the plan.
So the article's title is obviously sensationalized.
vidarh 49 minutes ago [-]
I have no problem believing that a Claude Max plan can consume equivalent to $5000 worth of retail Opus use, but one interesting thing you'll see if you e.g. have Claude write agents for you, is that it's pretty aggressive about setting agents to use Sonnet or even Haiku, so not only will most people not exhaust their plans, but a lot of people who do will do so in part using the cheaper models. When you then factor in Anthropics reported margins, and their ability to prioritise traffic (e.g. I'd assume that if their capacity is maxed out they'd throttle subscribers in favour of paid by the token? Maybe not, but it's what I'd do), I'd expect the real cost to them of a maximised plan to be much lower.
Also, while Opus certainly is a lot better than even the best Chinese models, when I max out my Claude plan, I make do with Kimi 2.5. When factoring in the re-run of changes because of the lower quality, I'd spend maybe 2x as much per unit of work I were to pay token prices for all my monthly use w/Kimi.
I'd still prefer Claude if the price comes down to 1x, as it's less hassle w/the harder changes, but their lead is effectively less than a year.
dimgl 5 hours ago [-]
Twitter.
gmerc 4 hours ago [-]
Nobody gets RSI typing “iterate until tests pass”
arthurcolle 3 hours ago [-]
Recursive self improvement and Repetitive Strain Injury being the same initialism is really funny to me
rs_rs_rs_rs_rs 2 hours ago [-]
Honest questions: have you never heard of a hyperbole before and are you on the spectum?
darkwater 1 hours ago [-]
Well, IDK, I have used CC with API billing pretty extensively and managed to spend ~$1000 in one month more or less. Moved to a Max 20x subscription and using it a bit less (I'm still scared) but not THAT less and I'm around 10% weekly usage. I'm not counting the tokens, though.
scuff3d 2 hours ago [-]
This article is hilariously flawed, and it takes all of 5 seconds of research to see why.
Alibaba is the primary comparison point made by the author, but it's a completely unsuitable comparison. Alibab is closer to AWS then Anthropic in terms of their business model. They make money selling infrastructure, not on inference. It's entirely possible they see inference as a loss leader, and are willing to offer it at cost or below to drive people into the platform.
We also have absolutely no idea if it's anywhere near comparable to Opus 4.6. The author is guessing.
So the articles primary argument is based on a comparison to a company who has an entirely different business model running a model that the author is just making wild guesses about.
simianwords 2 hours ago [-]
What? Aws is a good comparison if you want only infra level costs which is what the post is talking about.
beepbooptheory 4 hours ago [-]
Ok but so it does cost Cursor $5k per power-Cursor user?? Still seems pretty rough..
dimava 18 minutes ago [-]
Cursor may be losing money only on $200 sub people who do over $200 of usage (it grants $400)
Everyone else pays them at API prices
scriptsmith 4 hours ago [-]
Yes, you could turn it around to say that using Anthropic models in Cursor, Copilot, Junie, etc. is 'subsidising' Claude Code users.
arthurcolle 3 hours ago [-]
$5 = $5
but $5 that I amortize over 7 years might end up being $1.7 maybe if I don't rapidly combust (supply chain risk)
oefrha 3 hours ago [-]
No, to use $5k in Cursor you have to pay $5k.
unlimit 4 hours ago [-]
I wonder how they are defining a power user. How many tokens, what could be the size the code base?
dietr1ch 3 hours ago [-]
The $5k power user is the one that consistently uses all input and output tokens available under the Max subscription
fnord77 3 hours ago [-]
> I'm fairly confident the Forbes sources are confusing retail API prices with actual compute costs
Aren't they losing money on the retail API pricing, too?
> ... comparisons to artificially low priced Chinese providers...
Yeah, no this article does not pass the sniff test.
versteegen 3 hours ago [-]
> Aren't they losing money on the retail API pricing, too?
No, they aren't, and probably neither is anyone else offering API pricing. And Anthropic's API margins may be higher than anyone else.
For example, DeepSeek released numbers showing that R1 was served at approximately "a cost profit margin of 545%" (meaning 82% of revenue is profit), see my comment https://news.ycombinator.com/item?id=46663852
bandrami 2 hours ago [-]
Weird that they're all looking for outside money then
aurareturn 6 minutes ago [-]
Inference is profitable. No one is selling at a loss. It’s training to keep up with competitors that is causing losses.
vidarh 46 minutes ago [-]
They're all looking for outside money because they're all looking for outside money, and so need to keep up with their competitors investments in training. It's a game of chicken. Once their ability to raise more abates, they'll slow down new training runs, and fund that out of inference margins instead, but the first one to be forced to do so will risk losing market share.
Rendered at 08:44:26 GMT+0000 (Coordinated Universal Time) with Vercel.
And APIs are on-demand service equivalant.
Priority is set to APIs and leftover compute is used by Subscription Plans.
When there is no capacity, subscriptions are routed to Highly Quantized cheaper models behind the scenes.
Selling subscription makes it cheaper to run such inference otherwise many times your capacity is just sitting there idle.
I've used qwen code, codex and claude.
Codex is 2x better than Qwen code and Claude is 2x better than Codex.
So I'd hope the Claude Opus is atleast 4-5x more expensive to run than flagship Qwen Code model hosted by Alibaba.
This sloppy Forbes article has polluted the epistemic environment because now theres a source to point to as "evidence."
So yes this post author's estimation isn't perfect but it is far more rigorous than the original Forbes article which doesn't appear to even understand the difference between Anthropic's API costs and its compute costs.
It is not. It's a terrible comparison. Qwen, deepseek and other Chinese models are known for their 10x or even better efficiency compared to Anthropic's.
That's why the difference between open router prices and those official providers isn't that different. Plus who knows what open routed providers do in term quantization. They may be getting 100x better efficiency, thus the competitive price.
That being said not all users max out their plan, so it's not like each user costs anthropic 5,000 USD. The hemoragy would be so brutal they would be out of business in months
Opus isn't that expensive to host. Look at Amazon Bedrock's t/s numbers for Opus 4.5 vs other chinese models. They're around the same order of magnitude- which means that Opus has roughly the same amount of active params as the chinese models.
Also, you can select BF16 or Q8 providers on openrouter.
They do have different infrastructure / electricity costs and they might not run on nvidia hardware.
It's not just the models.
Namely, Amazon Bedrock and Google Vertex.
That means normalized infrastructure costs, normalized electricity costs, and normalized hardware performance. Normalized inference software stack, even (most likely). It's about a close of a 1 to 1 comparison as you can get.
Both Amazon and Google serve Opus at roughly ~1/2 the speed of the chinese models. Note that they are not incentivized to slow down the serving of Opus or the chinese models! So that tells you the ratio of active params for Opus and for the chinese models.
If Opus was 10x larger than the chinese models, then Google Vertex/Amazon Bedrock would serve it 10x slower than Deepseek/Kimi/etc.
That's not the case. They're in the same order of magnitude of speed.
These are not cell phone plans which the average joe takes, they are plans purchased with the explicit goal of software development.
I would guess that 99 out of every 100 plans are purchased with the explicit goal of maxing them out.
When I have a feeling that these tools will speed me up, I use them.
My client pays for a couple of these tools in an enterprise deal, and I suspect most of us on the team work like that.
If my goal was to max out every tool my client pays, I’d be working 24hrs a day and see no sunlight ever.
I guess it’s like the all you can eat buffet. Everybody eats a lot, but if you eat so much that you throw up and get sick, you are special.
Why? Because in my experience, the bottleneck is in shareholders approving new features, not my ability to dish out code.
if i hit the limit usually i'm not using it well and hunting around. if i'm using it right i'm basically gassed out trying to hit the limit to the max.
I find it a good comparison because it is a good baseline since we have zero insider knowledge of Anthropic. They give me an idea that a certain size of a model has a certain cost associated.
I don't buy the 10x efficiency thing: they are just lagging behind the performance of current SOTA models. They perform much worse than the current models while also costing much less - exactly what I would expect. Current Qwen models perform as good as Sonnet 3 I think. 2 years later when Chinese models catchup with enough distillation attacks, they would be as good as Sonnet 4.6 and still be profitable.
…You could take efficiency improvement rates from previous models releases (from x -> y) and assume; they have already made “improvements” internally. This is likely closer to what their real costs are.
Ask Opus to figure out how much it would cost. Lol.
Anthropic's models may be similar in parameter size to model's on open router, but none of the others are in the headlines nearly as much (especially recently) so the comparison is extremely flawed.
The argument in this article is like comparing the cost of a Rolex to a random brand of mechanical watch based on gear count.
Are Anthropic currently unable to sell subscriptions because they don’t have capacity?
Absolutely! Im currently paying $170 to google to use Opus in antigravity without limit in full agent mode, because I tried Anthropic $20 subscription and busted my limit within a single prompt. Im not gonna pay them $200 only to find out I hit the limit after 20 or even 50 prompts.
And after 2 more months my price is going to double to over $300, and I still have no intention of even trying the 20x Max plan, if its really just 20x more prompts than Pro.
They have a business model and are trying to capture more revenue, fully saturating your computer isn't obviously a good business strategy.
The entertainment industry. They still tell you about how much money they're leaving on the table because people pirate stuff.
What would happen in reality for entertainment is people would "consume" far less "content".
And what would happen in reality for Anthropic is people would start asking themselves if the unpredictability is worth the price. Or at best switch to pay as you go and use the API far less.
I think it's the other way around? Sparse use of GPU farms should be the more expensive thing. Full saturation means that we can exploit batching effects throughout.
I mean... rolex is overpriced brand whose cost to consumers is mainly just marketting in itself. Its production cost is nowhere close to selling price and looking at gears is fair way of evaluating that
When has production cost had anything to do with selling price?
Only thing that matters is if the users would have paid $5000 if they don't have option to buy subscription. And I highly doubt they would have.
I'm sure Anthropic is making money off the API but I highly doubt it's 90% profit margins.
Unlikely. Amazon Bedrock serves Opus at 120tokens/sec.
If you want to estimate "the actual price to serve Opus", a good rough estimate is to find the price max(Deepseek, Qwen, Kimi, GLM) and multiply it by 2-3. That would be a pretty close guess to actual inference cost for Opus.
It's impossible for Opus to be something like 10x the active params as the chinese models. My guess is something around 50-100b active params, 800-1600b total params. I can be off by a factor of ~2, but I know I am not off by a factor of 10.
Comparing tps ratios- by saying a model is roughly 2x faster or slower than another model- can tell you a lot about the active param count.
I won't say it'll tell you everything; I have no clue what optimizations Opus may have, which can range from native FP4 experts to spec decoding with MTP to whatever. But considering chinese models like Deepseek and GLM have MTP layers (no clue if Qwen 3.5 has MTP, I haven't checked since its release), and Kimi is native int4, I'm pretty confident that there is not a 10x difference between Opus and the chinese models. I would say there's roughly a 2x-3x difference between Opus 4.5/4.6 and the chinese models at most.
> Comparing tps ratios- by saying a model is roughly 2x faster or slower than another model- can tell you a lot about the active param count.
You sure about that? I thought you could shard between GPUs along layer boundaries during inference (but not training obviously). You just end up with an increasingly deep pipeline. So time to first token increases but aggregate tps also increases as you add additional hardware.
Hint: what's in the kv cache for the 2nd token?
And that's called layer parallelism (as opposed to tensor parallelism). It allows you to run larger models (pooling vram across gpus) but does not allow you to run models faster.
Tensor parallelism DOES allow you to run models faster across multiple GPUs, but you're limited to how fast you can synchronize the all-reduce. And in general, models would have the same boost on the same hardware- so the chinese models would have the same perf multiplier as Opus.
The Trillions of parameters claim is about the pretraining.
It’s most efficient in pre training to train the biggest models possible. You get sample efficiency increase for each parameter increase.
However those models end up very sparse and incredibly distillable.
And it’s way too expensive and slow to serve models that size so they are distilled down a lot.
I'd say Opus is roughly 2x to 3x the price of the top Chinese models to serve, in reality.
Of course, intense sparsification via MoE (and other techniques ;) ) lets total model size largely decouple from inference speed and cost (within the limit of world size via NVlink/TPU torrus caps)
So the real mystery, as always, is the actual parameter count of the activated head(s). You can do various speed benchmarks and TPS tracking across likely hardware fleets, and while an exact number is hard to compute, let me tell you, it is not 17B or anywhere in that particular OOM :)
Comparing Opus 4.6 or GPT 5.4 thinking or Gemini 3.1 pro to any sort Chinese model (on cost) is just totally disingenuous when China does NOT have Vera Rubin NVL72 GPUs or Ironwood V7 TPUs in any meaningful capacity, and is forced to target 8gpu Blackwell systems (and worse!) for deployment.
Opus is 2T-3T in size at most.
However, I'd say its relatively well assumed in realpolitik land that Chinese labs managed to acquire plenty of H100/200 clusters and even meaningful numbers of B200 systems semi-illicitly before the regulations and anti-smuggling measures really started to crack down.
This does somewhat beg the question of how nicely the closed source variants, of undisclosed parameter counts, fit within the 1.1tb of H200 or 1.5tb of B200 systems.
If you remove the cached token cost from pricing the overall api usage drops from around $5000 to $800 (or $200 per week) on the $200 max subscription. Still 4x cheaper over API, but not costing money either - if I had to guess it's break even as the compute is most likely going idle otherwise.
Why would it go idle? It would go to their next best use. At least they could help with model training or let their researchers run experiments etc.
Training currently requires nvidia's latest and greatest for the best models (they also use google TPU's now which are also technically the latest and greatest? However, they're more of a dual purpose than anything afaik so that would be a correct assesment in that case)
Inference can run on a hot potato if you really put your mind to it
I am not saying this would be a great use of their compute, but idle is far from the only alternative. (Unless electricity is the binding constraint?)
anthropic doesn't have that. single provider, single pricing decision. whether or not $5k is accurate the more interesting question is what happens to inference pricing when the supply side is genuinely open. we're seeing hints of it with open router but its still intermediated
not saying this solves anthropic's cost problem, just that the "what does inference actually cost" question gets a lot more interesting when providers are competing directly
Maybe the common factor here is not having deep/sufficient knowledge on the topic being discussed? For the article I mentioned, I feel like I was less focused on the strength of the writing and more on just understanding the content.
LLMs are very capable at simplifying concepts and meeting the reader at their level. Personally, I subscribe to the philosophy of - "if you couldn't be bothered to write it, I shouldn't bother to read it".
“what X actually is”
“the X reality check”
Overuse of “real” and “genuine”:
> The real story is actually in the article. … And the real issue for Cursor … They have real "brand awareness", and they are genuinely better than the cheaper open weights models - for now at least. It's a real conundrum for them.
> … - these are genuinely massive expenses that dwarf inference costs.
This style just screams “Claude” to me.
It has enough tells in the correct frequency for me to consider it more than 50% generated.
Popular content is popular because it is above the threshold for average detection.
In a better world, platforms would empower defenders, by granting skilled human noticers flagging priority, and by adopting basic classifiers like Pangram.
Unfortunately, mainstream platforms have thus far not demonstrated strong interest in banning AI slop. This site in particular has actually taken moderation actions to unflag AI slop, in certain occasions...
1. It would be nice to define terms like RSI or at least link to a definition.
2. I found the graph difficult to read. It's a computer font that is made to look hand-drawn and it's a bit low resolution. With some googling I'm guessing the words in parentheses are the clouds the model is running on. You could make that a bit more clear.
I thought there was no moat in AI? Even being 10x costlier, Anthropic still doesn't have enough compute to meet demand.
Those "AI has no moat" opinions are going to be so wrong so soon.
So no, Claude would not be getting NEARLY as much usage as it's currently getting if it weren't for the $100/$200 monthly subscription. You're comparing Kimi to the price that most people aren't paying.
[1] https://www.wheresyoured.at/anthropic-is-bleeding-out/ [2] https://www.wheresyoured.at/costs/
> My LinkedIn and Twitter feeds are full of screenshots from the recent Forbes article on Cursor claiming that Anthropic's $200/month Claude Code Max plan can consume $5,000 in compute.
So the article's title is obviously sensationalized.
Also, while Opus certainly is a lot better than even the best Chinese models, when I max out my Claude plan, I make do with Kimi 2.5. When factoring in the re-run of changes because of the lower quality, I'd spend maybe 2x as much per unit of work I were to pay token prices for all my monthly use w/Kimi.
I'd still prefer Claude if the price comes down to 1x, as it's less hassle w/the harder changes, but their lead is effectively less than a year.
Alibaba is the primary comparison point made by the author, but it's a completely unsuitable comparison. Alibab is closer to AWS then Anthropic in terms of their business model. They make money selling infrastructure, not on inference. It's entirely possible they see inference as a loss leader, and are willing to offer it at cost or below to drive people into the platform.
We also have absolutely no idea if it's anywhere near comparable to Opus 4.6. The author is guessing.
So the articles primary argument is based on a comparison to a company who has an entirely different business model running a model that the author is just making wild guesses about.
Everyone else pays them at API prices
but $5 that I amortize over 7 years might end up being $1.7 maybe if I don't rapidly combust (supply chain risk)
Aren't they losing money on the retail API pricing, too?
> ... comparisons to artificially low priced Chinese providers...
Yeah, no this article does not pass the sniff test.
No, they aren't, and probably neither is anyone else offering API pricing. And Anthropic's API margins may be higher than anyone else.
For example, DeepSeek released numbers showing that R1 was served at approximately "a cost profit margin of 545%" (meaning 82% of revenue is profit), see my comment https://news.ycombinator.com/item?id=46663852