Their valuations differ by about 13%. That's close enough that I wouldn't call it "blown past".
Things change fast in this space. Anthropic had a big boost from having the premier coding model for a while, but GPT-5.5 has closed that gap at a time when a lot of Anthropic customers are looking for cheaper alternatives.
Anthropic is coming off of a recent change to their enterprise billing that substantially changed the pricing for many users. They were smart to do the fundraising before the effects of that change could fully propagate.
The acceleration rate has been extraordinary… they went from mostly unknown outside AI circles to the number one player almost overnight. If that’s not “blown past” I don’t know what is.
The branding of Claude is so much stronger than ChatGPT. Even Anthropic is such better branding than OpenAI (especially considering they're not open at all).
My wife knows about Claude because that's what I use and we pay for. She uses it also as a result. And inevitably she will talk about Claude to her friends.
If "normie" means a noncorporate knowledge worker who uses the free version, yes.
For enterprise, Anthropic is crushing it. In the manufacturing sector I anecdotally hear a 2:1 ratio of Claude to ChatGPT for teams who are settling on a platform.
At my company the grassroots advocacy from devs has certainly been for Claude Code.
Unfortunately even though we have a degree or two of seperation from most federal contracts the punitive DoD blacklisting had enough of a chilling effect on our legal team to make them drag their feet on approving any contract involving Anthropic.
So I pitched OpenAI Business with Codex so we could drop our Github Copilot Business subscription before the billing change takes effect June 1st which was approved without pushback.
I felt some responsibility for finding an immediate solution to dump Copilot since I was the one who recommended adopting Copilot in the first place, ugh... Our prices would have quadrupled based on the single month Microsoft in their beneficence allowed previewing with their tool to simulate what the post-rug pull pricing would have looked like.
Codex becoming more or less a 1:1 replacement for CC made that a no brainer given our options and the exploitative value proposition of Copilot under the new pricing model (which Microsoft evidently hoped companies like us would just accept despite being a third tier option in the dev space these days).
ChatGPT is a word now. People may use Perplexity, or Google, or Grok to ask questions online. And later they tell you "ChatGPT told me this". It's a new "I googled in Yahoo".
ChatGPT is the 5th most visited site (as well as has nearly a billion weekly active users) and none of the competitors are even close. In the consumer space, Gemini is doing well but Claude is not even in the same galaxy. OpenAI is undoubtedly the leader in consumer LLMs and by a large margin. I'm sure there are mixups, but if someone is telling you they're using chatGPT, they almost certainly mean they're using chatGPT.
The consumer market is worthless though. Consumers will never pay, so the only revenue option is ads which barely, if even at all, pay for inference costs.
Ads implemented remotely competently would be worth a lot of money and more than pay for inference. Inference is cheap, especially outside token expensive ordeals like agentic coding.
Maybe. To really make money on ads they would need to embed them directly into the chat I think. Banner ads arent worth enough I think and google is able to make so much off them largely because people are already looking to click a link when they search something. People would just ignore them with genAI.
Maybe Im projecting my distaste for being psychologically manipulated, but I dont think users would continue using a genAI that embeds ads directly into the response when they can just switch to gemini where they only see banners.
I’m pretty sure Gemini would be the leader in consumer LLMs considering it’s on every single search result. Every single google search is also usage of gemini.
Google stuffing things in the search results of existing users does not mean active participation or usage. (Not that I'm saying it's not getting used but it's just a feature of google search, and only a fraction of the kind of queries llms get anyway)
People use LLMs for a lot of things. Different kind of search is only one of them. AI mode is not stopping people from using ChatGPT because it's just a subset of consumer LLM queries.
Perhaps it's just regional, but I've been noticing more and more people saying "chat" to describe ANY ai chat interface including ChatGPT. They might have a Kleenex problem on their hands.
>Few non-programmers have heard of anthropic or claude
They ran a super bowl ad. It's all over the construction industry. Claude is still not quite the Kleenex that ChatGPT is, but there is a pretty good chance lay people have heard of ChatGPT, Gemini, and Claude by now.
To disagree with the person below/above me that ChatGPT is the word used generically, when someone uses Gemini or Claude or Copilot, they TELL you which one they used, because they are essentially saying "i didnt use ChatGPT by choice."
Gemini is the one most likely to be used without people knowing which one they used.
Def sounds like a bubble to me. In my own bubble, ChatGTP is so well known over the others that people will often slip and refer to other AI services collectively as ChatGTP.
e.g. "I put it in chatgtp and..." when they actual asked Gemini.
It sounds like you have no clue what you're talking about.
The people in real life who say ChatGBT or similar are either so out of the loop the technology doesn't even matter to them, or simply are just stupid.
There is no way a person who can't get "GPT" right is even worth listening to.
My sister, who isn't in tech and would be called a "normie" by people more online than myself, told me she switched to Claude a few months ago because of Anthropic's fight with the pentagon. IMO the unbubbled public certainly knows about Anthropic/Claude, especially given their Super Bowl ad and their stance on standing up to the pentagon/Trump admin.
Ironically their tussle with the US federal government is what made them a household name [1]
There's no better way to create awareness of a brand than to get it featured in the most popular reality TV show globally at the moment: "Thing Trump Did: Season 2."
Comparing $/MTokfor models makes as much sense as comparing $/ghz for CPUs. Models have different tokenizers and take varying number of "thinking" to get to a solution. A far better proxy is how much it takes to do a run, which takes all of that into account. Such metrics are much harder to gather, but once source claims $3357 for gpt-5.5 vs $4686 for opus, the opposite of your conclusion.
There is no conclusion , I only stated the only objective fact to compare with that will not change for you to me.
Everything else is subjective to your setup, use case, configuration tuning and so forth.
More importantly bean-counters and decision makers at even 150+ seat orgs are looking at pricing sheets and enterprise contracts not how it performs for some team in a specific harness today to make million dollar annual contracts. It is not common for procurement teams to do commission the level of detailed analysis or large scale pilots that will actually hold for the duration of contract.
That doesn't mean that GPT-5.5 is selling less than Claude at all, just that cost is not the primary driver if list price is not cheaper, there is reason these are published in the same format by every vendor, because the common metric is how finance likes to compare with.
Most variants of GPT-5.5 are less chatty and token-intensive than Opus 4.8/4.7, so despite the output token price being higher, it generates fewer tokens, so the net cost is lower.
Per-token pricing is totally sensible from the provider-perspective on mapping COGS to revenue, but for a consumer, different models will produce more or less tokens, meaning the cost calculation is multi-dimensional.
You can configure model to be terse/concise with output style ? There are plenty of popular projects like https://github.com/JuliusBrussee/caveman which do it for you even.
Input/Cache/Output ratios are use case and configuration dependent . Any benefits in one model can usually be roughly to another with configuration tuning, and discussions devolve into subjective experience.
Pricing sheet is the objective way to compare cost.
Anthropic is at the mercy of 3rd party datacenter contracts. AFAIK OpenAI will soon run mostly on on their own GPUs.
I don't like Altman and I am still upset about his memory deal last year but he prepared for the current shortages months before anybody else. Meanwhile, Anthropic seems to lack any plans besides third party contracting. IMHO they got very lucky with xAI and Google having spare capacity and willing to rent it. But what about next year?
Which also leaves OpenAI vulnerable to NVidia's aggressive pricing. To my knowledge Anthropic is relatively well positioned across multiple compute vendors/hardware providers.
It also leaves OpenAI vulnerable to any GPU breakthroughs. You could imagine company X comes up with a XPU that is 100% faster than what's currently there.*
We are still in the short-half-life phase of GPUs. If a 2x faster GPU is on the horizon, why wouldn't OpenAI already be in line to buy? They aren't buying just 1, they are buying multiple datacenters' worth. So they wouldn't be a low priority, back of the line customer.
A short half-life means you are going to quickly dispose of what you have now, anyway. In fact most current datacenters can't even handle Vera Rubin, so I don't think there's short term risk here.
Nvidia has probably monopolized several upstream supplies to manufacture critical chip components for next 2 years, the HBMs and Optics component from LITE, as well as TSM capacity. Let alone those power components they funded themselves.
Let's say you have a genius design, but you will have it close to impossible to compete with Nvidia in getting it to volumes.
Jensen is a player, he isn't fooling around with all these Asian trips just to wine and dine
nVidia can only 'monopolize' these components for itself inasmuch as other industry players are not seriously interested in them. This can change rather quickly.
Everyone has critical risk on multiple parts of the supply chain. GPUs and Memory are just things OAI mitigated for.
Power - Bigger bottleneck than GPU or RAM perhaps, New Grid connected capacity is typically 10+ year timescale with lot of regulatory friction. Captive capacity is also quite constrained - now Gas turbines have 7+ year wait time.
There are plenty of hard constraints that OAI cannot easily solve either.
Stargate as a project is real, they only stoped the Stargate UK thing.
Anthropics relativ longterm contract with xAI def shows that they can fill the capacity vs Musk not. OpenAI and Anthropic are both using a lot of capacity so its fair to say that this is an advantage.
If they stay very close competitive (which they are), your own datacenter does reduce token price.
>Anthropic is at the mercy of 3rd party datacenter contracts
I mean, this is a bit like complaining that McDonalds doesn't have their own herds of cows. OpenAI actually isn't in the business of buying GPUs or running data centres, and it's pretty weird to think that's an advantage (though it comes up constantly on here, as Anthropic keeps eating OpenAI's lunch).
There are many suppliers that are desperate to fight for Anthropics business, and it has shown an agility to embrace whatever advances in the industry come along. Anthropic is now running across a million or so Google TPUv8s, for instance. If tomorrow someone else comes out with a better GPU/TPU, they can embrace it in a heartbeat.
All while OpenAI sits on their rapidly depreciating GPUs.
Or...actually they won't, because OpenAI doesn't take business advice from HN. The vast majority of OpenAI's compute is from Microsoft, Oracle and so on. They're smart enough to not become a big hardware purchaser when that isn't their business. The core claim of your comment simply isn't true at all, nor is that the direction OpenAI is moving.
Anthropic is riding a hype wave as a result of brilliant marketing. OpenAI has the better products, higher reliability and better community relations. I don't expect the situation to continue.
I disagree. They have been winning lately because of better harnesses and interfaces. New actual decent features are shipped almost weekly on Claude code and Claude desktop.
I’m not so sure. We only need to look at Uber’s example of companies realizing they’re spending way too much and trying to rein it in. Claude has excellent revenue but it is highly dependent on very rich technology companies continuing to spend lavishly without seeing returns. The music will stop at some point and Anthropic will be hit the hardest. OpenAI may have less revenue but it is distributed across many, many more customers and use cases, it’s resilient. And even if Anthropic do, somehow, manage to keep their customers spending huge amounts on Claude, they’re very vulnerable to being undercut by OpenAI given codex is pretty much at parity. Anthropic seems more vulnerable to me.
I think it's somewhat guaranteed that the music will at least die down a little bit. We saw this with cloud companies being bitten by cloud cost optimization initiatives. I can't imagine we won't see the same with AI, especially as the workforce stops trying to tokenmaxx to save their role.
Every week there's at least one post on the HN front page bitching about API errors from Claude because Anthropic doesn't have enough serving capacity. I really don't see any signs they're "spending too much", the actual evidence on the ground seems to be exactly the opposite: constant exasperation that they're not spending enough.
I just finished talking to a dev manager friend of mine at a household name company.
He told me they are massively pulling back on the AI stuff.
Right now the lashback is about cost, because that's the most easily measured pain point.
Soon, we'll start seeing a deeper understanding of the quality issues. At that point, it's likely this whole experiment gets firmly put in a bin of the toolbox where it belongs.
I know people at medium size companies where they are tracking AI costs very carefully. They are pulling back to levels under $100/week in AI spend per engineer, encouraging use of lower quality, lower cost models, etc.
$100/week will get you both a higher tier of Claude and significantly more tokens with GLM5.1 or Kimi, both of which are competitive in a decent harness (for Kimi that means their own CLI - it works well in that, but has a lot of quirks that requires special treatment). Just slightly more will get you all three.
I don't doubt you, but $100 is approximately the cost to company of one hour of dev time. If companies end up being willing to spend only 2% of their dev budget on AI, this bubble is not going to last long.
I mean Anthropic’s customers are spending too much on Claude. Anthropic’s customers are encouraging tokenmaxxing amongst their employees; measuring employees by token usage. That’s great for Anthropic’s short term revenue numbers but terrible long term because at some point companies will realize tokenmaxxing is not good. OpenAI is much less exposed to tokenmaxxing, which is a good thing.
Tokenmaxxing is the practice of measuring employees by how many tokens they use, encouraging employees to burn tokens needlessly, it is unrelated to what agents can do.
If a task can be completed with 100k tokens but employees are considered better performers if they complete it with 500k tokens instead… that’s unsustainable and cannot possibly benefit Anthropic in the long term.
At some point, Amazon and Uber and so on and so forth are going to realize that actually, employees using 100k tokens or even 50k tokens is better than 500k and Anthropic’s revenue will fall off a cliff.
I think removing limits is fine. There’ll be overspend and at some point adjustments in expectations as we learn more about the value that can be delivered which will likely result in a reduction in spend, but even now, during this period of relative immaturity about measuring the value of output, so long as more tokens = more output, I don’t think the introduction of limits represents much of a risk to Anthropic and OpenAI. Tokenmaxxing is uniquely bad because it is not tied to any additional value (more tokens for the same output).
And I could be wrong about tokenmaxxing being a Claude specific problem but as far as I can tell, all of the major companies encouraging employees to maximize their token usage are Claude Code users. And the music has to stop on that at some point, whether because the companies run out of money or because they learn better ways of measuring productivity in the AI age. And if tokenmaxxing is what is driving Anthropic’s lead in revenue, it could be catastrophic to lose that, because Anthropic are spending billions of dollars per month on the infrastructure to support it.
If tokenmaxxing is evenly distributed between Anthropic and OpenAI then they’ll both hurt but equal hurt shouldn’t disadvantage either much.
This business and financial race is probably the craziest in human history, so zig-zags are expected. One company may take advantage on one curve while another is stuck in the pits.
OpenAI isn't shaky or vulnerable, this market will need at least 2 players.
I see most of the surge here comes FOMO AI spending which will have to be dialed down later half of the year, otherwise those companies will have to layoff to fund their AI bill, which is harmful to their business.
Anthropic grabs its bag at the peak, but feast is over.
How? OpenAI and Antrophic are basically the Big 2 racing away at light speed; the others who can't get near them are perhaps shaky & vulnerable. And sure, there's a garden full of those.
Because the market almost certainly can’t support two foundation model labs given the increasingly little difference across models and the massive sums of cash required to keep it all going. There is no big 2, just a race to survive and be the big 1.
It probably can't support any because there's no moat and smaller, open source models are catching up. This is like investing $1T into mainframe computers in 1980.
There's at least two markets here. Consumer ad driven and worker augmentation markets. Likely a 3rd as a backend infrastructure provider to a bunch of value add companies.
I think Google has caught up enough to certainly be a player in the consumer ad driven market.
I also don't think only one foundation model adds up. Now that the trail is blazed a dozen companies can likely make a good enough model. The question is if there's a moat to make it winner take all
Google needs to catch up on what? Devs mindshare? The latest Opus 4.8 carefully selected benchmarks made sure to pick Gemini 3.1 Pro and not Gemini 3.5 Flash: 3.5 Flash is beating Opus 4.8 on several of the benchmarks Anthropic posted but simply was ignored.
I don't think SOTA-wise Google has a lot of catch up to do.
Gemini 3.5 Flash is not good at coding in practice. Gemini 3.1 Pro too, in particular is known to be bad at tool calls. Many companies would love to have alternatives to Claude Code (as it's a significant risk to depend on one vendor), so far most of the buzz is about moving to Codex but much fewer talk about moving to Gemini. All these benchmarks are not very informative, the Chinese labs do better on these benchmarks than in practice, for example.
Yeah it's very murky, but if they're arguably profitable under some definition of profitable, then it's ridiculous to claim there's "no path to profitability"
Especially if you assume that 6 months ago they weren't very close to this version of profitable
In terms of personal assistant AI the monopolists have a massive advantage because they control the platforms and can box out competitors from deeply integrating with the OS.
Data. Google has access to unphasmable amount of real human-created data with zero expectations of privacy (wink wink Apple): videos, photos, search, navigation, mobile app usage including competition platforms, emails, etc.
Both Anthropic and OpenAI only has access to whatever they can buy or steal.
And it's becoming increasingly hard to get fresh uncontaminated data for training. No amount of money can buy that.
> Both Anthropic and OpenAI only has access to whatever they can buy or steal.
A trillion can buy you quite a lot! Like offer some company a ton of money for data, and if they say no simply buy said company. Bonus points if it's someone like Atlassian who's stock price is getting hammered largely because of you.
