To people asking why, this is a good lesson on the Collison’s ambitions. Stripe is one of the best API companies in the world. They know how to serve high volumes of latency and availability sensitive requests. They’ve abstracted the financial rails for payments and now want to abstract the rails for LLMs.
They’re the perfect company to own OpenRouter.
Tokens are simply a lightweight valuable asset. Stripe can serve as the middleman as well as anyone. They know how to route to many providers (payment rails) with huge differences in service characteristics. LLM providers are far easier.
Then they can work this into an offering where users can subscribe to tokens and use them across services. It solves one of the core monetization challenges of every AI company: how do you price when your costs are variable on usage, but nobody can make sense of charging by token?
From here, they can start hosting their own models and competing as an AWS for tokens. They can be the best provider of $OPEN_MODEL, or their own, and optimize for you.
Here are some things that are abundantly clear to me.
1.) LLMs are useful for programming
2.) Open models are excellent and will continue to improve
3.) Economies of scale and ease of access mean self hosting is out of the question for a large number of users
This means that even if the largest labs are not worth trillions and a large amount of the data center build out is not as valuable as the builders project and GPU/RAM prices plummet, it will not matter at all for this business. People will want to buy cheap open source tokens from a centralized trusted provider.
$7 billion for a business with little overhead that is already within their core competency and has strategic growth potential seems like a very good deal.
Unfortunately we don't really know anything about OpenRouter or Stripe's numbers. They're both private companies. AFAIK there really isn't much information publicly available about how profitable or not OpenRouter actually is, or whether it has a little or a lot of overhead. So I can't really say whether $7B is reasonable or inflated. It seems like an enormous number, and my hunch is like all the other enormous numbers floating around in the AI space it's probably inflated by the hype bubble. But we just don't know.
Broadly I agree with you. It seems like they have an in-demand product and there could be a sustainable business there at least in principle. But whether it's a $7B business or a $70M business I can't say.
What does openrouter actually do though? They buy a bunch of tokens from token providers and allow you to easily route queries between token processors?
This seems like the kind of thing that you can have Claude write in an afternoon for whatever service you're running. I don't see the value.
My understanding is that it’s a common LLM API with common billing for multiple LLM providers. You can switch between LLM providers whenever you like, and the way you pay for API usage doesn’t change. You don’t even need to create an account with the LLM provider.
I don’t use OpenRouter, but exe.dev has a similar feature. I choose which LLM to use in a pulldown menu.
But the downside of this is that you’re paying API prices, which are much higher than subscription prices.
So, recently they added a way to connect a ChatGPT subscription and I mostly use that. But I can try out other models any time.
So this doesn’t seem like all that much of a moat to me, but it’s still a convenient service to have, since you aren’t going to set up billing with all the LLM providers in an afternoon. I expect there will be multiple competitive businesses.
Compare with what distributors do for physical goods.
Why a centralized trusted provider? Linux VMs are also useful (host services that are the end result of programming) and open source and fit for cloud convenience. I dont go to OpenLinuxVM to get my vms. I just get them from whereever. Aws. Digital ocean. You name it. Why centralize tokens?
depends what the tulips are. tokens? private growth stage shares?
openrouter is involved with a lot of scammy crypto personalities. maybe they are the tulips people have been manic about for too long. maybe the collisons are the tulips.
Ok, at first I didn’t get it and thought it would make sense if stripe just wants to build payments for agents, but on a second thought your idea about “once buy tokens — use everywhere” is very good one!
I mean. They used to be good. When they started out their docs were absolutely far above the competition. They aren’t now, and Stripe has lived long enough that they’ve seen themselves become the villain.
Feature bloat definitely made their API and product docs harder than the early days. Stripe Checkout would have saved my teams days, but I agree, they need to go back to their roots and make the 99% use case very easy.
Stripe's APIs are very good. They could be better, but it's difficult to change the underpinnings of financial infrastructure. A lot of risk, little incentive to move, and you're designing something that has to last (and make sense) for 15+ years in the future.
I worked on this at Stripe in 2022. We were the first teams to start building v2 APIs and data models to solve exactly this problem. The first target launch date (in Feb 2022) was November 2022. It was launched in May of 2025.
My reporting line, as an EM, was Netflix, Oracle, Oracle. No one had startup experience. It was drenched in politics. The engineers were largely brilliant, kind, and hardworking.
I still love the company and believe in Patrick. Believe me, he deeply understands what you're saying and wants it to be the best it can be. But it was clear to me, even then, that they'd lost a lot of what made them special. They could maintain it, but I wasn't sure they could do it again. Banking-as-a-Service was one opportunity, Link was another, and now this will be a third. We'll see. (I say this with a lot of love for Stripe and Stripes.)
Coincidentally, I had a conversation with a recruiter at Anthropic and saw them doing something very similar. They were starting a new team in a new vertical and wanted someone with experience running an org of 100+ people. I would bet real money that it will be a fraction of the product/impact it could be (though still probably make money!)
They are certainly much better than the competition. That said, do not believe the marketing hype. If you want to accept various kinds of payments, you will eventually run into the stupid limitations of the payment system, and Stripe APIs will not abstract this for you in spite of what they advertise.
For example, redirect-based payment methods (iDEAL, Bancontact, Sofort) are a huge source of complexity, and handling all the additional states a payment can be in is far from obvious. Stripe mostly just drops them in your lap and says "deal with it".
I work for a competitor and can confidently say they have amazing docs. Payments are complex, and they definitely do cover the bulk 80% case in a REALLY good way.
We have this workshop thing we make most new joiners go through where they integrate our own APIs into a demo app and the number 1 response from anyone familiar with the wider payments industry is that we should aim to make our docs more like Stripe's lol
You're asking how someone missed llms. On this website? That demonized them for years and is only now fighting the cultural battle to accept them as useful?
(Not picking on you here, you just provided a well-written peg for a popular narrative. I'm aiming to sharpen my own thinking here & perhaps learn something.)
This seems like a stretch given the rise of local inference, especially the Prism Labs rumors from a few weeks ago.
One way I think about LLMs is they are akin to fancy databases in that they are software of which you can ask questions and get answers if you ask properly. Oracle & SQL Server are akin to OpenAI and Anthropic, and there are analogues for MySQL, PostgreSQL, SQLite, MongoDB, PlanetScale, etc. (This is an analogy, it's not going to be a perfect fit.)
In that view, would it make sense for someone to say that their credit card processor just bought the company that makes their ODBC driver? Would anybody suggest that the TPS of their RDS instance is a lightweight asset?
I don't see this as strategic beyond the obvious idea that Stripe wants to get closer to AI, and they haven't been able to get the market to care about their natural linkage to AI (Radar).
> they can start hosting their own models and competing as an AWS for tokens
The financial pressure of doing this has caused cuts to core product teams at the richest companies (which Stripe isn't!). I would not want my payments processor to go down this road and to get worse at processing payments.
> would it make sense for someone to say that their credit card processor just bought the company that makes their ODBC driver? Would anybody suggest that the TPS of their RDS instance is a lightweight asset?
Would is make sense to say that their online bookstore now sells Ethernet cables, bidets, and delivers groceries? Sells _cloud infrastructure_??
Amazon made two transitions:
1. Amazon the online bookseller => Amazon "The Everything Store"
2. Amazon the online retailer => Amazon the Cloud Services company
I believe what Stripe is doing here is closer to (1). "We are good at high-throughput APIs that wrap complexity with thin margins. We did it for credit cards, then ~all payment methods, now other digital bits."
In this context, tokens are much, much easier than international payments.
Now: Is it outlandish for their customers?
Not really. Every engineer knows stripe as an engineering company. I don't think Stripe is what it once was, but it's certainly a generational company. You're asking engineers—who broadly have a positive impression of Stripe—to use this product they already know. OpenRouter gets the positive brand association (trust) of Stripe; Stripe expands into a new domain whose technical needs are extremely similar.
re: the Amazon transitions, the first shareholder letter lays out the plan to go beyond books. As I remember it, books were always only supposed to be the entry point.
AWS transition was more around building the platform Amazon.com needed to grow, and also to monetize the same platform.
Neither of these really fit with Stripe.
Stripe can obviously operate OpenRouter, they have the tech skills. The risk to the core business is that OpenRouter's growth path will distract from the core financial business and/or require a very different capital stack. (Someone already suggested Stripe scale out first-party model running, which can get very expensive.)
Your simplification itself is compressing away the meaningful point, and then you're confused about your own over-simplification.
Stripe is not a "credit card processor", and that's not the point of OP's comment. The point is that being the intermediary between merchants and processors is the valuable expertise in terms of an OpenRouter acquisition.
It's been a few years, but I used to work on almost exactly this (not at Stripe, but we processed hundreds of billions annually). If you're a big enough merchant (or Stripe itself), you can choose to send a credit card transaction to any one of potentially many processors, each of which have their own performance characteristics (one may approve at a higher rate, the other may charge better fees, etc). All of this subject to attributes of the transaction itself (ticket size, geography, card type, many others). You know quite a bit about the transaction itself before you send it out, so you can build up routing knowledge to optimize for whatever thing you care about (usually transaction success rates or fees).
See how this starts to look a lot like OpenRouter with money instead of tokens? I'm not sure I 100% believe that's how it'll shake out, but there is a transferable skillset.
I think the "money instead of tokens" is the important part. Money and tokens are fundamentally different was part of my argument (that I did not make well). It's not clear that it is a good thing for money movers to get into the token business (this apples to Ramp as well).
Banks are also intermediaries between parties (at scale, really between any parties). One could use similar logic to say that it therefore makes sense for Wells Fargo to start buying homebuilders because the home buyers will eventually be paying Wells anyway.
BankRate similarly processes volume of consumer mortgage quotes in real-time communication with lender APIs. They also do not have a reason to own a token router, even though their business involves similar processes to what you outline for Stripe.
Being an intermediary in a value chain does not mean you're critical path, or that it makes sense for you to be in the critical path.
Funnily enough, you argue that this is a natural fit for Stripe while a peer reply argues that it's a change in the business a la Amazon->AWS.
> The point is that being the intermediary [..] is the valuable expertise
Yeah, all those intermediary businesses must be eliminated first and foremost if you want a better internet space. The current state of affairs is the direct product of just too many intermediaries sucking money out of both ends that could be spent on quality of service, which leads to lower prices, which leads to a healthier market overall.
I don't think your analogy is a good one. Yes, in some way an LLM is a like a Db, but with a fundamental difference: the data is part of the database engine. So when you are the middleman like Stripe/OpenRouter in this case, you are in the middle of a stream of data, which has value on its own - since people pay for it - and some specific characteristics (cost, speed, "intelligence", latency etc) based on which a router can take decisions.
You'd be surprised how little volume of api calls payments companies get compared to advertising companies or consumer internet companies. Stripe aren't especially well placed to own this asset on a technical level. That doesn't make this a bad acquisition per se, but there is no magic stripe can inject here as far as infra/technology is concerned.
Lago cloud ? Probably not , lago like sentry is more likely to be self-hosted .
Stripe completed the metronome acquisition their usage billing offering just 8months back , that hardly makes Stripe the largest player in this place when they are not even fully integrated .
Would be great if either one would be able to back their claims with some data. But yes, I'd also imagine that SaaS version gets more traffic, but maybe there are some self-hosted enterprise editions that get insane amount of traffic
Yes exactly and once you factor in that a lot of sites have the Stripe JS code on every single page the total number of requests is quite high... Plus they do outbound webhook requests for dozens if not hundreds of different events so a single end to end customer checkout actually can fire dozens of webhook events (depending on what is set up for the particular merchant - it does for us!)
That’s volume in terms of total payments, but in terms of number of API requests it’s surprisingly little.
I used to work in payments myself and I was shocked by how little the throughout is in terms of actual transactions per second. There simply aren’t that many payments happening globally at any moment, especially when compared to other industries.
A single games company can be doing far more API requests than the total volume of payments happening concurrently in the world, ditto for ads companies or observability companies.
At a payments company, 500 payments per second is considered high throughput and 1000 payments per second is a huge amount of volume. I’ve worked at games companies where 5000 transactions per second is low throughput.
The hard part in payments companies is consistency and reliability and “correctness”, the actual load and volume is embarrassingly small.
You'd be surprised how antiquainted many of them are. I don't mean Stripe or Braintree but the long tail of terrible terrible credit card processors. If I hear "Chase Payment Gateway" I shudder.
More expensive API calls though, I imagine. It’s also not that small scale. Sure, it’s no Google, but certainly higher scale than most companies using them
Yea, I totally get why they want to be in this space (especially if you see transaction volume becoming more agentic), but why buy it? Why not do what Cloudflare or Vercel or AWS, etc did and just build your own?
Stripe already has killer APIs and knows how to build API infrastructure at Extreme scale, it still has developer cred after all these years, they are already Kings of multiplexing to different sources -- so why not just build your own LLM gateway?
My only guesses are: they want to buy the traffic instead of trying to organically grow it (but I feel like Stripe would get the volume anyway because of name rec), they want to buy the revenue and think they can dramatically reduce COGs because of efficiencies, buying the relationships with the frontier labs (who all dogfood their LLMs in pseudo-private on OR), or... something else?
A lot of the freakout and "bye" posts on the OpenCodeCLI subreddit due to DeepSeek raising prices.
Stripe is a middleman. So is OpenRouter. So is OpenCode. Unless you own the data center and the hardware, how cheap can a middleman's tokens really be compared to the hyperscalers, without massive model compression?
Even DeepSeek itself is raising token prices. How much margin is there for a middleman like Stripe buying tokens in bulk from a data center and reselling them? Last i check Stripe do not run or own physical data center
That's old economy thinking. If you have shit tons of data about the behavior of large groups of people, someone will pay you for it. HFT firms paying for satellite pictures of shopping mall parking lots in order to get an advantage over their competitors is old news, but it gives you an idea of the game being played. Why do you think meta bought gif keyboard?
I guess everyone is selling tulips now. For me, if this deal goes through, I'll find an OpenRouter competitor. I have no interest in Stripe backing this. I realize: I don't matter. But I'll continue to vote with dollars. Eventually the masses will matter again.
Why should businesses trust a MITM that can snoop on their LLM conversations? It’s one thing to share confidential data with a model provider if that’s the only way to get the benefit, but given all the hand wringing I see from executives and infosec folks about restricting who internal information can be shared with, I don’t see why they’d let additional parties have access to it as well.
It’s like we spent all this time enabling ubiquitous TLS to keep enable end-to-end encryption of all communications, but all of a sudden we’re now totally okay with having a 3P observe and record all of them so we can save a few bucks.
Stripe operates in a highly regulated environment and have earned the trust of the banks, merchants and payment networks. But LLMs are operating in a completely unregulated environment and I’m not sure the trust Stripe has earned for the former is transitive to the latter.
I wonder if this deal is primarily just to buy payment volume.
OpenAI just announced earlier this week that Ayden would become their payment provider (when it was previously Stripe).
And OpenRouter has a large percentage of overall AI payment volume for all the major labs.
Both OpenAI and OpenRouter represent ~$100B in payment volume, whereas Stripe in total doing ~$2T. Two customer doing ~5% of your total volume who didn’t even exist a few years ago, must be kind of scary for Stripe.
Stripe would lock in volume but how exactly are they monetizing that? Any margin the get from OpenRouter is money extracted from OpenRouter’s revenue (which is now theirs.)
7 billion one off for 100 billion annually? If I am understanding that correctly it feels like a no brainer to me…
Also if you think about it differently… OpenRouter is adjacent to what Stripe is but for getting access to AI models. If they dont mess this up they could invest in openerouter and grow that 100b pie.
Adyen just bought Orb [0], a usage-based billing company, for $335M, specifically so they could position themselves as the payments choice for AI companies.
OpenAI is a large company, and large companies go multi processor for payments. I would check the press release carefully, and see if you can confirm that Adyen is taking all of OpenAI's volume, or even a majority of it.
How can a middle man for api calls be worth so much? Their market share can’t be very large right? For comparison, $7B is more than market cap of Lyft, Dolby, and Alaska Airlines. What is happening?
If your data is leaking to OpenRouter, how many companies would be OK with that? So I doubt OpenRouter are peeking inside the workloads that flow through their proxy. However, they do publish "Top Models by Task" ranking[1] so maybe you have a point.
the providers you route to will of course have their own policies, which openrouter surfaces to you through the webui and api. you can even configure automatic routing to select providers based on your policy preferences.
Exactly , not even small startup anywhere in us , eu or nearby will be ok to send data to China, but with open router you can create a no China provider selector with one llm call and start using Chinese llms without paying to vendors a single cent .
What’s the angle for stripe , electrify over tokens exchange is the new money flow , and stripe wants to monetize it. 5% tax on any llm token is an amazing deal
For one, openai and anthropic are the top two AI labs and therefore have some reputation at stake. Openrouter is like aliexpress. Any fly-by night vendor can get listed on there.
Good point, the American labs have an awful reputation, and public sentiment is so low it might as well be negative. Plus they've already broken our trust by scraping and stealing everyone's intellectual property to begin with, so they have a lot of catching up to do in terms of reputation and trust.
>Good point, the American labs have an awful reputation ...
Reddit/bluesky users seething at anthropic/openai doesn't matter, because they're not the one making the purchasing decisions for AI. Moreover all the complaints you mentioned applies to chinese AI labs as well, with the extra issue that they're beholden to the CCP. Taking the side of chinese ai labs because they're the competitors of american labs is "enemy of my enemy is my friend" logic, which is just dumb. It's like those people who think the US is bad, so then go and simp for china/russia.
Openrouter is very transparent about where your requests are going. Every single one is logged.
You decide the routing if you want.
I trust it because there are at least 7 billion on the line. If it came out that they were violating their contract and sending company secrets to China when they promised they won’t, they would lose literal billions for basically no gain, and possible criminal charges.
Why on earth would you think they are sending data to China after you click that toggle when they have every reason not to?
As a QA engineer with over 30 years of identifying software regressions, I would assume that data would eventually get sent to China even if you click that toggle. One ambitious intern or new hire is all it takes, especially in an age of agent-generated code and 1K+ line PRs.
With that logic you shouldnt even deploy to the cloud at all. Someone in AWS can also accidentally share an S3 bucket with backups to the whole internet, there are plenty of precedents for that
It's an inflated number due to the AI market, but their selling point is distribution.
Many people and businesses want to experiment with different models, but they don't want to sign up for a dozen different services. Businesses can make it difficult to approve new vendors. If your company is looking at 5 different vendors for tokens and teams can't agree to switch together, OpenRouter comes along with a unified interface and a single billing point.
They also become the point to add value-add services on top in a portable way. They already offer some things like automatic JSON repair, but I can see them adding functionality like leak detection tools, monitoring, alerts, and other patterns that a company can set up once and use with all of the models theirs teams need.
What I'm not so sure about is their moat. They have the brand recognition, but it seems rather easy for someone else to build what they've built. I'm a little confused about why Stripe didn't just build the same thing internally. Acquiring this company gives them an instant boost of 10 million customers for their AI business, which might be key to some financial goal they've got.
>It's an inflated number due to the AI market, but their selling point is distribution.
>Many people and businesses want to experiment with different models, but they don't want to sign up for a dozen different services. Businesses can make it difficult to approve new vendors. If your company is looking at 5 different vendors for tokens and teams can't agree to switch together, OpenRouter comes along with a unified interface and a single billing point.
Isn't that ripe for being picked off by aws or azure? Both already have marketplaces where vendors can offer whatever cloud services they want. Both already offer first party inference service, and have contracts with all the stodgy corps where it's "difficult to approve new vendors". Not to mention they have IAM and SSO built in. Good luck bolting that onto a third party vendor like openrouter.
AWS Bedrock doesn’t even try to stay on top of new open source models or their API capabilities. We’ve also run into also kinds of frustrating issues rate limiting and slowness issues. GCP is probably the best alternative now that they have actual spending caps, I believe a lot of Openrouter models are run on GCP.
The other popular option is something like LiteLLM, which just has a major vulnerability that left a lot of big corps exposed. After spending a year trying to fight this battle, I’ve decided OpenRouter is worth their cut (and maybe more but don’t tell them).
As a nobody I like OpenRouter because it lets me quickly test out all of the emerging open weight models. I normally use Claude Code but if I run out of usage I'll pull up GLM 5.2 or Deepseek V4 in OpenCode via OpenRouter. Occasionally I'll run Claude's work by a Google/OpenAI model through OpenRouter. If Claude Code becomes too expensive I now know I'll be pretty comfortable with open weight models.
You can't really compare market caps like that. Addressable market or market share is only one piece of the puzzle, and the other companies you mention are in very different kinds of markets, with very different kinds of products, with very different kinds of costs and margins.
A valuation just reflects what someone thinks about the future cash flows of the business.
But yeah, it does feel a bit crazy; unsurprisingly, AI hype affects valuations of AI companies too. On the other hand, I can see the idea that some people might be betting on the idea that the big US labs are bloated and spend too much money, and that the real money is going to be in serving open-weight models, and/or in automatically combining and routing to different models based on the task at hand.
The market cap of any business operating in the physical realm has to account for enormous liability, especially for one like an airline. Also, their operating costs are huge with much less ability to scale, resulting in much lower potential increase in margins.
I prefer using OpenRouter because the money I pay them in premium is far less than the money I lose by not using it on other providers. On OpenRouter, I can switch dollars between models. But for the providers, after evaluating I'm stuck with some number of credits on ones I don't use. This kind of intermediation is actually super useful to me. I must imagine that they can also provide other things that are of value like provider quality measures, ability to swap providers during downtime, etc.
You could just as easily ask yourself how can a middle man for people finding web pages be worth so much, but here we are, and Google is worth something like $4 trillion, mostly on the back of search.
As it turns out, it's very valuable to be the intermediary between a large number of people who want something and a commoditized market of providers. It's the middleman who captures most of the margin.
Now, AI models are not a commodity yet. But things seem like they might be heading that direction. And in a world where they are, Stripe probably wants to be that guy sitting in the middle.
It's pretty valuable to own the customer touch point
Also when people ask questions like this it usually means they're asking questions about a point in time, eg given today's numbers
But valuation should account for trajectory (where will they be in five years?)
In the hypothesized bull case for ai, they benefit dramatically from the secular tailwinds (ie unrelated to their own business strategy) of AI
So one way to consider this is it's a hedge by stripe in case AI becomes as big as some people think it is. And 7b to get in on the ai wave is a good deal in that case. And in all the other cases, it goes to zero and stripe is fine
OpenRouter raised money at a $1.3 billion valuation a few months ago, if the NYTimes reported valuation is accurate.
Going from a $1.3b valuation to a $7b exit in a couple months is an amazing return for those investors. I hope the OpenRouter employees got some decent equity out of this
Historically acquisitions have never really been good for customers. Time for me to look for an OpenRouter alternative? At least they're also as easy to switch from as the model providers they proxy.
One reason is to avoid a potential boiling frog scenario. I'm already in one with Android, but there are no viable - for me - alternatives (never touching Apple and mobile Linux isn't there yet) so I'm truly stuck. If there's a chance things may start changing at OpenRouter in such a way that months/years from now I'll look back and say "oh, how I wish I'd switched providers", I'd rather switch now just out of principle than be boiled.
Any implication for the privacy and confidentiality of our queries? Openrouter yes sees everything. But as long as they are independent and establishing their core business on privacy and confidentiality reputation, I feel more comfortable with this than with a new broker. AI queries can contain more sensitive information than just financials. And financial informations are protected by law, btw. AI queries, however, are not.
Stripe seems to appear to be attempting to compete with the US Federal Reserve as the core dollar payment processor while also dipping into the same for tokens. Core value exchange platform, dollars, tokens, whatever. A bit of a less exciting version of Dune. “The transaction volume must flow.”
Perhaps the US State Dept. used it's strong relationship with Stripe to encourage this purchase as a geopolitical move. They've done so before with companies like eBay and Microsoft to neuter the P2P telephone access network that was "Skype".
Both are in the business of putting a single API key in front of a fragmented ecosystem and charging a convenience fee. This middleman business is naturally coalescing.
The nature of the ecosystem also means that pricing is closely tied to "procurement" which could be routing, limits, whatever at a company level.
If stripe wants to be _the_ one that charges that fee, they either have to continuously try to ensure that all the different middlenen use stripe (most of them do!) but even better is to acquire the largest middleman.
You don't want someone else who happens to do all the routing+limits+policies, end up not using stripe. They already have this hold in existing stripe financial products where they apply all the policies, and everything goes through them.
It is also an easy deal from an investor point of view.
As a payments processor, Stripe has conversion data. They can route based on ROI with end-to-end A/B testing or online optimization. There's a tricky credit assignment / multi-touch attribution problem to solve for some use cases, too, that could benefit from transfer learning across their customers. Also, the (unstructured) data they collect from OpenRouter users can be used to improve many other dimensions of their business, including fraud detection. Lastly, both businesses look fairly similar in unit economics if you squint hard enough; for example, they both collect a ~3-5% transaction fee from most customers.
The strategy here is that the same services that Stripe offers for payments have corollaries in the LLM world. Security, user management, perhaps injection attack monitoring, etc.
However I think they dearly overpaid for this as the core technology behind Stripe (fraud detection and integration with global banks) is hard to replicate. Even with the features mentioned above I think the technologies behind OpenRouter are vastly easier to replicate, perhaps even trivial now.
This kind of makes a lot of sense. While my first stake was surprise that a proxy is valued at such a high rate, But it really comes down to the terms if open router is allowed by their terms to see the prompts and responses, Then that's very, very valuable data today for any model improvements and other opportunities that people might be looking for.
I'm surprised OpenRouter went for 7B, while fal.ai just raised at $8b, despite fal having far less traffic (semrush reports ~5x more traffic to openrouter). They seem like very parallel businesses, just with focuses on different models (creative models vs LLMs).
If someone from Stripe / OpenRouter is listening, please let me provide Open Router OAuth to my users and let me take a cut of their usage. Apple takes 30, I would take whatever the users are willing to pay. 5 to 15 is reasonable.
I'm still amazed at Apple's 30%, it's just so brazenly high (like: "fuck you, we're Apple, we do what we want and you'll lick it up"). Unfortunately, too, other companies follow Apple's lead, and so this massive rent seeking percentage has become normalised.
I can totally agree that Apple sucks for doing this, but you seem to insinuate they started this. As always Apple wasn’t first not even tenth: Nintendo, Steam, Xbox, Kindle all did it prior to “the platform economy” 3)
Didn't know there was prior art of that 30% scale, as I generally avoid as much of that shit as I can, Apple is the one I hear about the most.
I amend my prior comments (with actually being able to amend them) to Apple sucks as much as the rest of the rent seekers, and in reference to Aeolun's comment, it doubly sucks that enough people choose to lick it up that the rent seekers don't feel ashamed of their rent seekery, but in fact profit handsomely from it, thus encouraging others to follow suit.
It feels ridiculous that you can't buy directly from the software publisher - like the old internet when they were called programs. I feel I may be on the end of pointing and laughing at the old man, but the situation now is literally a gatekeeping middle-man that gets a 30% cut. WTF? Nice work if you can get it.
Makes me feel like a rube (in a significant majority, mind) for the few games I've bought off Steam. Eesh, even GOG is 30%. Epic's seems reasonable at 12% and only after the first million in revenue per year.
Software publishers need a 'donate' button on their websites where erstwhile pirates can cough launder cough their procurements.
What value are you providing that you'd take some further cut? You could take the money from the user via stripe and then pay your own openrouter bill instead.
I’ve stopped using OpenRouter this year. And I stopped using different models for testing. Anthropic won. I’ve even let my OpenAI API credits expire (which is a ridiculous policy in itself) because I had no need for them.
I vividly remember back in 2023/2024, I was sitting on my sofa thinking, "there has to be a market to abstract data access layer, something like Terraform but for LLM provider to avoid vendor lockin". Only if I had access to capital to build something like that at that time, I could have been my try for an AI start up.
I don't really understand the value -add, so far "smart" routing currently produces worse results at higher cost and static routing can be done yourself trivially. Plus the risk of sending the most sensitive data you have to a third party.
Everyone seems to be asking why this a great strategic fit for Stripe.
Stripe can now provide tools to every product that sells metered AI usage and take a cut. This isn't about extracting a small percent on the tokens flowing from your coding agents to your model of choice. It's about all the products that are going to come to market and monetize metered usage.
Want to analyze your 2026 tax return? Use VisorAI's Tax Agent and pay only for what you use. Stripe provides all the accounting, payment processing, ships money to vendors, and takes a bit off the top. Trillions of dollars moving from the labor market to the token market? It's $10B per point in fees for every trillion.
The reason OpenRouter can build value even if there are 100 clones is switching costs and flexibility. AWS Bedrock may be great for large enterprises, but flexible startups will use OpenRouter for ease, and now they have Stripe's distribution. Once you use OpenRouter, you won't switch because you become embedded in the logs and cost-saving systems.
A large missing link for AWS Bedrock is Gemini model support [0]. And this is a vital feature, because Gemini itself (as far as I'm aware) does not have a real-time daily or weekly hard spend cap in the case of misconfigured or compromised callers [1][2].
Meta's closed source models, too, are only available on OpenRouter (and it's touted as a first-class recommendation on [3]).
So, in many ways, OpenRouter is the only game in town to access some of the highest-funded model families on the market through a single API endpoint.
I'm unsure why AWS is ceding this territory (there have to be ways to allow users to opt into the lesser governance inherent in using third party providers, while not diminishing the brand's governance requirements). But, since they are, it's a meaningful opportunity left open.
Well, you have the logs and specific cost control measures and setups in place. You would lose all that in the event you attempted to switch, plus Stripe owning OpenRouter gives them stronger distribution and further solidifies the switching costs in the broader integrated solution Stripe offers - https://s-1.vercel.app/posts/what-stripe-can-become-broader-...
yeah, i bet this seems like a good deal in a few years, we'll see. remember Poe haha that was a similar kinda thing but openrouter waited until the market was mature
Free model usage is controlled by the LLM provider, not OpenRouter. Both parties have their own incentives to allow it and that would not change with different owners.
Terrible news for the market. Even as someone who dislikes AI, it's terrible. Stripe is a horrible company and more market consolidation is very worrisome.
Hope someone clones it and they take a loss at it.
They do reach out pretty quick when your usage takes a huge dive after you move to one of the single provider plans that offer hard to believe possibly unsustainable value.
I had consistent problems (5 months back) with OpenRouter and other providers. And very hard to debug. I used the provider directly (Google) and worked perfectly.
Why would PayPal acquiring OpenRouter make it any more of a monopoly than it is? Not saying it is or isn't, but OpenRouter isn't even a financial company.
That's not a very common pattern, to be honest. A bunch of people have been experimenting with automatic model routing recently but mainly as a cost optimization, since tokens for the best models have got expensive once you start piping millions of tokens through them.
I haven't seen much evidence that model routing is being widely used yet. I think it's still more of an experimental mechanism right now.
OpenRouter is a proxy, not an automatic router. Rather than building API clients for five different AI providers, you build one client to OpenRouter, and switching models become extremely easy. This matters when new models are coming out virtually every day.
All the model companies except kind of Anthropic (and even they half-assedly do) implement the OpenAI API. It's not an open standard but, like the S3 API, it effectively is.
And, to answer your question, no. The existence of a common API makes it trivial to change zero code and send requests to a different model.
The Anthropic messages API is a competing standard (it's just better than the OpenAI API which even OpenAI has moved away from) and some Chinese providers use it as their standard.
Isn't this what LiteLLM is doing? And is Open Source? Maybe I am asking a dumb question because this is the age-old SaaS vs OSS debate, but I am struggling to find the angle here.
no. i don't know if openrouter even guarantees consistent protocl across models.
the value of openrouter is it offers centralized billing. you can route your calls to any provider you want, test a whole bunch of models against each other, and you just get one bill from openrouter. switching to a new model, or a new provider of the same model, doesn't mean setting up a new billing account with a new provider.
I see, that make sense, thanks for the explanation. Because otherwise I was thinking their whole business model would be just to provide an API tool library.
It's more complicated than that. Lots of providers use an "OpenAI-ish" API, but many of them have subtle differences in things like tool calling or thinking blocks. OpenRouter normalizes the wire format.
OpenRouter does more than just proxying; they also aggregate providers for open-weight models, which has a stabilizing effect on pricing and gives you protection against a single provider's downtime.
Dumb question but aren’t they all OpenAI API compliant? I thought that was pretty standard, eg I know Anthropic works with it. Also Claude Code can work with different models, they’re probably not using OpenRouter for that?
Effectively but not quite. Anthropic does mostly support the OpenAI API but there are minor feature gaps. You have to use Anthropic's bespoke API to use everything (which may or may not be important to you).
There are some gotchas with it too. e.g. The "system" role message isn't supported in the Anthropic API. It's not the biggest deal but it's a potential footgun.
On the other hand, the openrouter SDK + API takes care of this for you: Their one gateway has ways of enabling provider-specific features. You can switch the model name and pretty much be good to go.
Claude Code does not work with other models out of the box. You need a wrapper around Claude Code that translates other model requests + respondes to what the harness expects.
The APIs are similar but not the same, which is to say: different. Definitely not as simple as swapping out the base URL.
Claude Code has no built-in way to use non-Anthropic models. It does support overriding the API URL with an environment variable, but there are drawbacks. OpenRouter is not used within Claude Code at all.
I think that OpenRouter's goal is to have the user pick the model, so there's manual configuration for the user control and cost benefit. Maybe you meant OpenCode Zen? But let's be honest, "AI" is definitely incredible, not "hilarious", or "so bad".
To people asking why, this is a good lesson on the Collison’s ambitions. Stripe is one of the best API companies in the world. They know how to serve high volumes of latency and availability sensitive requests. They’ve abstracted the financial rails for payments and now want to abstract the rails for LLMs.
They’re the perfect company to own OpenRouter.
Tokens are simply a lightweight valuable asset. Stripe can serve as the middleman as well as anyone. They know how to route to many providers (payment rails) with huge differences in service characteristics. LLM providers are far easier.
Then they can work this into an offering where users can subscribe to tokens and use them across services. It solves one of the core monetization challenges of every AI company: how do you price when your costs are variable on usage, but nobody can make sense of charging by token?
From here, they can start hosting their own models and competing as an AWS for tokens. They can be the best provider of $OPEN_MODEL, or their own, and optimize for you.
unless it's all just tulip mania, then they just spent $7B on nothing
Here are some things that are abundantly clear to me.
1.) LLMs are useful for programming
2.) Open models are excellent and will continue to improve
3.) Economies of scale and ease of access mean self hosting is out of the question for a large number of users
This means that even if the largest labs are not worth trillions and a large amount of the data center build out is not as valuable as the builders project and GPU/RAM prices plummet, it will not matter at all for this business. People will want to buy cheap open source tokens from a centralized trusted provider.
$7 billion for a business with little overhead that is already within their core competency and has strategic growth potential seems like a very good deal.
Unfortunately we don't really know anything about OpenRouter or Stripe's numbers. They're both private companies. AFAIK there really isn't much information publicly available about how profitable or not OpenRouter actually is, or whether it has a little or a lot of overhead. So I can't really say whether $7B is reasonable or inflated. It seems like an enormous number, and my hunch is like all the other enormous numbers floating around in the AI space it's probably inflated by the hype bubble. But we just don't know.
Broadly I agree with you. It seems like they have an in-demand product and there could be a sustainable business there at least in principle. But whether it's a $7B business or a $70M business I can't say.
What does openrouter actually do though? They buy a bunch of tokens from token providers and allow you to easily route queries between token processors?
This seems like the kind of thing that you can have Claude write in an afternoon for whatever service you're running. I don't see the value.
people are lazy
people don't want to maintain infra (ie adding new models all the time)
people want their queries to work without thinking
if a provider goes down, openrouter queries dont (ideally)
it's really not that complex to understand
i think plenty of people don't want to be forced to using just Claude and ChatGPT
My understanding is that it’s a common LLM API with common billing for multiple LLM providers. You can switch between LLM providers whenever you like, and the way you pay for API usage doesn’t change. You don’t even need to create an account with the LLM provider.
I don’t use OpenRouter, but exe.dev has a similar feature. I choose which LLM to use in a pulldown menu.
But the downside of this is that you’re paying API prices, which are much higher than subscription prices.
So, recently they added a way to connect a ChatGPT subscription and I mostly use that. But I can try out other models any time.
So this doesn’t seem like all that much of a moat to me, but it’s still a convenient service to have, since you aren’t going to set up billing with all the LLM providers in an afternoon. I expect there will be multiple competitive businesses.
Compare with what distributors do for physical goods.
Why a centralized trusted provider? Linux VMs are also useful (host services that are the end result of programming) and open source and fit for cloud convenience. I dont go to OpenLinuxVM to get my vms. I just get them from whereever. Aws. Digital ocean. You name it. Why centralize tokens?
depends what the tulips are. tokens? private growth stage shares?
openrouter is involved with a lot of scammy crypto personalities. maybe they are the tulips people have been manic about for too long. maybe the collisons are the tulips.
Ok, at first I didn’t get it and thought it would make sense if stripe just wants to build payments for agents, but on a second thought your idea about “once buy tokens — use everywhere” is very good one!
Do people feel Stripe APIs are really that good? I’ve always found them unintuitive, tons of options, weird edge cases, breaking occasionally, etc.
I’m sure payments are convoluted, but I’d still imagine they could be meaningfully easier for the bulk 80% of use case?
Yes, Stripe is extremely good at what they do. Payments are complicated.
Payments themselves are so much complicated. Payments to support multi-currency, jurisdictions, countries, scenarios, that's complicated.
I'm friends with a payment docs writer. Holy crap are payments complicated.
I have implemented the Swift payment Spec and I can tell you having this abstracted in a nice API is so much better.
I mean. They used to be good. When they started out their docs were absolutely far above the competition. They aren’t now, and Stripe has lived long enough that they’ve seen themselves become the villain.
Feature bloat definitely made their API and product docs harder than the early days. Stripe Checkout would have saved my teams days, but I agree, they need to go back to their roots and make the 99% use case very easy.
Stripe has the worst payment APIs, except for all the others.
I expect many of the warts in the API surface are there because they want to avoid changing things for fear of breaking someone else's business.
Stripe's APIs are very good. They could be better, but it's difficult to change the underpinnings of financial infrastructure. A lot of risk, little incentive to move, and you're designing something that has to last (and make sense) for 15+ years in the future.
I worked on this at Stripe in 2022. We were the first teams to start building v2 APIs and data models to solve exactly this problem. The first target launch date (in Feb 2022) was November 2022. It was launched in May of 2025.
My reporting line, as an EM, was Netflix, Oracle, Oracle. No one had startup experience. It was drenched in politics. The engineers were largely brilliant, kind, and hardworking.
I still love the company and believe in Patrick. Believe me, he deeply understands what you're saying and wants it to be the best it can be. But it was clear to me, even then, that they'd lost a lot of what made them special. They could maintain it, but I wasn't sure they could do it again. Banking-as-a-Service was one opportunity, Link was another, and now this will be a third. We'll see. (I say this with a lot of love for Stripe and Stripes.)
Coincidentally, I had a conversation with a recruiter at Anthropic and saw them doing something very similar. They were starting a new team in a new vertical and wanted someone with experience running an org of 100+ people. I would bet real money that it will be a fraction of the product/impact it could be (though still probably make money!)
> Netflix, Oracle, Oracle
Well that brings back some memories. I remember there was a third Oracle in that chain, but he left earlier compared to the others.
They are certainly much better than the competition. That said, do not believe the marketing hype. If you want to accept various kinds of payments, you will eventually run into the stupid limitations of the payment system, and Stripe APIs will not abstract this for you in spite of what they advertise.
For example, redirect-based payment methods (iDEAL, Bancontact, Sofort) are a huge source of complexity, and handling all the additional states a payment can be in is far from obvious. Stripe mostly just drops them in your lap and says "deal with it".
I work for a competitor and can confidently say they have amazing docs. Payments are complex, and they definitely do cover the bulk 80% case in a REALLY good way.
We have this workshop thing we make most new joiners go through where they integrate our own APIs into a demo app and the number 1 response from anyone familiar with the wider payments industry is that we should aim to make our docs more like Stripe's lol
why did they miss the market that is OpenRouter?
You're asking how someone missed llms. On this website? That demonized them for years and is only now fighting the cultural battle to accept them as useful?
> Tokens are simply a lightweight valuable asset
(Not picking on you here, you just provided a well-written peg for a popular narrative. I'm aiming to sharpen my own thinking here & perhaps learn something.)
This seems like a stretch given the rise of local inference, especially the Prism Labs rumors from a few weeks ago.
One way I think about LLMs is they are akin to fancy databases in that they are software of which you can ask questions and get answers if you ask properly. Oracle & SQL Server are akin to OpenAI and Anthropic, and there are analogues for MySQL, PostgreSQL, SQLite, MongoDB, PlanetScale, etc. (This is an analogy, it's not going to be a perfect fit.)
In that view, would it make sense for someone to say that their credit card processor just bought the company that makes their ODBC driver? Would anybody suggest that the TPS of their RDS instance is a lightweight asset?
I don't see this as strategic beyond the obvious idea that Stripe wants to get closer to AI, and they haven't been able to get the market to care about their natural linkage to AI (Radar).
> they can start hosting their own models and competing as an AWS for tokens
The financial pressure of doing this has caused cuts to core product teams at the richest companies (which Stripe isn't!). I would not want my payments processor to go down this road and to get worse at processing payments.
> would it make sense for someone to say that their credit card processor just bought the company that makes their ODBC driver? Would anybody suggest that the TPS of their RDS instance is a lightweight asset?
Would is make sense to say that their online bookstore now sells Ethernet cables, bidets, and delivers groceries? Sells _cloud infrastructure_??
Amazon made two transitions:
1. Amazon the online bookseller => Amazon "The Everything Store"
2. Amazon the online retailer => Amazon the Cloud Services company
I believe what Stripe is doing here is closer to (1). "We are good at high-throughput APIs that wrap complexity with thin margins. We did it for credit cards, then ~all payment methods, now other digital bits."
In this context, tokens are much, much easier than international payments.
Now: Is it outlandish for their customers?
Not really. Every engineer knows stripe as an engineering company. I don't think Stripe is what it once was, but it's certainly a generational company. You're asking engineers—who broadly have a positive impression of Stripe—to use this product they already know. OpenRouter gets the positive brand association (trust) of Stripe; Stripe expands into a new domain whose technical needs are extremely similar.
Appreciate your thoughts.
re: the Amazon transitions, the first shareholder letter lays out the plan to go beyond books. As I remember it, books were always only supposed to be the entry point.
AWS transition was more around building the platform Amazon.com needed to grow, and also to monetize the same platform.
Neither of these really fit with Stripe.
Stripe can obviously operate OpenRouter, they have the tech skills. The risk to the core business is that OpenRouter's growth path will distract from the core financial business and/or require a very different capital stack. (Someone already suggested Stripe scale out first-party model running, which can get very expensive.)
Appreciate your taking the time to respond.
Your simplification itself is compressing away the meaningful point, and then you're confused about your own over-simplification.
Stripe is not a "credit card processor", and that's not the point of OP's comment. The point is that being the intermediary between merchants and processors is the valuable expertise in terms of an OpenRouter acquisition.
It's been a few years, but I used to work on almost exactly this (not at Stripe, but we processed hundreds of billions annually). If you're a big enough merchant (or Stripe itself), you can choose to send a credit card transaction to any one of potentially many processors, each of which have their own performance characteristics (one may approve at a higher rate, the other may charge better fees, etc). All of this subject to attributes of the transaction itself (ticket size, geography, card type, many others). You know quite a bit about the transaction itself before you send it out, so you can build up routing knowledge to optimize for whatever thing you care about (usually transaction success rates or fees).
See how this starts to look a lot like OpenRouter with money instead of tokens? I'm not sure I 100% believe that's how it'll shake out, but there is a transferable skillset.
I hear your argument, I do.
I think the "money instead of tokens" is the important part. Money and tokens are fundamentally different was part of my argument (that I did not make well). It's not clear that it is a good thing for money movers to get into the token business (this apples to Ramp as well).
Banks are also intermediaries between parties (at scale, really between any parties). One could use similar logic to say that it therefore makes sense for Wells Fargo to start buying homebuilders because the home buyers will eventually be paying Wells anyway.
BankRate similarly processes volume of consumer mortgage quotes in real-time communication with lender APIs. They also do not have a reason to own a token router, even though their business involves similar processes to what you outline for Stripe.
Being an intermediary in a value chain does not mean you're critical path, or that it makes sense for you to be in the critical path.
Funnily enough, you argue that this is a natural fit for Stripe while a peer reply argues that it's a change in the business a la Amazon->AWS.
Anyway, I appreciate your thoughts.
> The point is that being the intermediary [..] is the valuable expertise
Yeah, all those intermediary businesses must be eliminated first and foremost if you want a better internet space. The current state of affairs is the direct product of just too many intermediaries sucking money out of both ends that could be spent on quality of service, which leads to lower prices, which leads to a healthier market overall.
I don't think your analogy is a good one. Yes, in some way an LLM is a like a Db, but with a fundamental difference: the data is part of the database engine. So when you are the middleman like Stripe/OpenRouter in this case, you are in the middle of a stream of data, which has value on its own - since people pay for it - and some specific characteristics (cost, speed, "intelligence", latency etc) based on which a router can take decisions.
You'd be surprised how little volume of api calls payments companies get compared to advertising companies or consumer internet companies. Stripe aren't especially well placed to own this asset on a technical level. That doesn't make this a bad acquisition per se, but there is no magic stripe can inject here as far as infra/technology is concerned.
Agree, even in the payments space usage-billing metrics providers like Chargebee / Lago etc get order(s) of magnitude more data than payment.
Granted payments and associated ledges require a level of consistency that other systems do not require, but that is orthogonal to high volume.
Stripe also supports usage based billing... If you think Lago gets more volume than Stripe I just don't know what to tell you lol
Lago cloud ? Probably not , lago like sentry is more likely to be self-hosted .
Stripe completed the metronome acquisition their usage billing offering just 8months back , that hardly makes Stripe the largest player in this place when they are not even fully integrated .
Lol there is no way that self-hosted Sentry gets more volume than the SaaS version
You are way off here
Would be great if either one would be able to back their claims with some data. But yes, I'd also imagine that SaaS version gets more traffic, but maybe there are some self-hosted enterprise editions that get insane amount of traffic
Stripe did $1.9T payments in 2025. That's 1.6% of global GDP.
It's take a lot to persuade me that there is another payment provider close to that volume.
Lago did around $10B in total API payment volume for comparison.
Yes exactly and once you factor in that a lot of sites have the Stripe JS code on every single page the total number of requests is quite high... Plus they do outbound webhook requests for dozens if not hundreds of different events so a single end to end customer checkout actually can fire dozens of webhook events (depending on what is set up for the particular merchant - it does for us!)
> once you factor in that a lot of sites have the Stripe JS code on every single page
That's rarely a hit to origin though? That's mostly cached at the browser, if not at the CDN.
That’s volume in terms of total payments, but in terms of number of API requests it’s surprisingly little.
I used to work in payments myself and I was shocked by how little the throughout is in terms of actual transactions per second. There simply aren’t that many payments happening globally at any moment, especially when compared to other industries.
A single games company can be doing far more API requests than the total volume of payments happening concurrently in the world, ditto for ads companies or observability companies.
At a payments company, 500 payments per second is considered high throughput and 1000 payments per second is a huge amount of volume. I’ve worked at games companies where 5000 transactions per second is low throughput.
The hard part in payments companies is consistency and reliability and “correctness”, the actual load and volume is embarrassingly small.
You'd be surprised how antiquainted many of them are. I don't mean Stripe or Braintree but the long tail of terrible terrible credit card processors. If I hear "Chase Payment Gateway" I shudder.
More expensive API calls though, I imagine. It’s also not that small scale. Sure, it’s no Google, but certainly higher scale than most companies using them
Yea, I totally get why they want to be in this space (especially if you see transaction volume becoming more agentic), but why buy it? Why not do what Cloudflare or Vercel or AWS, etc did and just build your own?
Stripe already has killer APIs and knows how to build API infrastructure at Extreme scale, it still has developer cred after all these years, they are already Kings of multiplexing to different sources -- so why not just build your own LLM gateway?
My only guesses are: they want to buy the traffic instead of trying to organically grow it (but I feel like Stripe would get the volume anyway because of name rec), they want to buy the revenue and think they can dramatically reduce COGs because of efficiencies, buying the relationships with the frontier labs (who all dogfood their LLMs in pseudo-private on OR), or... something else?
That's the part I'm confused about.
A lot of the freakout and "bye" posts on the OpenCodeCLI subreddit due to DeepSeek raising prices.
Stripe is a middleman. So is OpenRouter. So is OpenCode. Unless you own the data center and the hardware, how cheap can a middleman's tokens really be compared to the hyperscalers, without massive model compression?
Even DeepSeek itself is raising token prices. How much margin is there for a middleman like Stripe buying tokens in bulk from a data center and reselling them? Last i check Stripe do not run or own physical data center
I doubt Stripe can do much better here.
That's old economy thinking. If you have shit tons of data about the behavior of large groups of people, someone will pay you for it. HFT firms paying for satellite pictures of shopping mall parking lots in order to get an advantage over their competitors is old news, but it gives you an idea of the game being played. Why do you think meta bought gif keyboard?
I guess everyone is selling tulips now. For me, if this deal goes through, I'll find an OpenRouter competitor. I have no interest in Stripe backing this. I realize: I don't matter. But I'll continue to vote with dollars. Eventually the masses will matter again.
what are you looking for in a competitor?
doing research for my end to end encrypted router TrustedRouter
Thanks for preemptively answering the question of why. And it makes sense.
Why should businesses trust a MITM that can snoop on their LLM conversations? It’s one thing to share confidential data with a model provider if that’s the only way to get the benefit, but given all the hand wringing I see from executives and infosec folks about restricting who internal information can be shared with, I don’t see why they’d let additional parties have access to it as well.
It’s like we spent all this time enabling ubiquitous TLS to keep enable end-to-end encryption of all communications, but all of a sudden we’re now totally okay with having a 3P observe and record all of them so we can save a few bucks.
Stripe operates in a highly regulated environment and have earned the trust of the banks, merchants and payment networks. But LLMs are operating in a completely unregulated environment and I’m not sure the trust Stripe has earned for the former is transitive to the latter.
But 7 Billion to start that process? That is an awful lot of money
shares. its paper. they issue new shares to buy the company. yes, its dilution, but its not the same as cash.
This makes no sense to me, payments are low volume traffic, their biggest constraints are getting business logic, security and transactions right.
LLMs are a different beast.
I wonder if this deal is primarily just to buy payment volume.
OpenAI just announced earlier this week that Ayden would become their payment provider (when it was previously Stripe).
And OpenRouter has a large percentage of overall AI payment volume for all the major labs.
Both OpenAI and OpenRouter represent ~$100B in payment volume, whereas Stripe in total doing ~$2T. Two customer doing ~5% of your total volume who didn’t even exist a few years ago, must be kind of scary for Stripe.
https://www.reuters.com/business/retail-consumer/rise-ai-sho...
https://stripe.com/newsroom/news/stripe-2025-update
Stripe would lock in volume but how exactly are they monetizing that? Any margin the get from OpenRouter is money extracted from OpenRouter’s revenue (which is now theirs.)
7 billion one off for 100 billion annually? If I am understanding that correctly it feels like a no brainer to me…
Also if you think about it differently… OpenRouter is adjacent to what Stripe is but for getting access to AI models. If they dont mess this up they could invest in openerouter and grow that 100b pie.
$100b is payment volume, not revenue. Stripe gross revenue from those payments would be on the order of tenths of a percent of that.
Plus the platform fee that OpenRouter makes? 5 billion a year?
I think volume is itself good and help Stripe negotiate lower rate with banks etc.
>OpenRouter represent ~$100B in payment volume, whereas Stripe in total doing ~$2T.
That is a much higher % than i expected.
> Two customer doing ~5% of your total volume who didn’t even exist a few years ago, must be kind of scary for Stripe.
Shopify uses Stripe no? Probably good volume discount though.
Adyen*
Adyen just bought Orb [0], a usage-based billing company, for $335M, specifically so they could position themselves as the payments choice for AI companies.
[0] https://www.withorb.com/
OpenAI is a large company, and large companies go multi processor for payments. I would check the press release carefully, and see if you can confirm that Adyen is taking all of OpenAI's volume, or even a majority of it.
Maybe it might also have to do with AI agent wallets and getting deeper in to the agent-based transactional business?
How can a middle man for api calls be worth so much? Their market share can’t be very large right? For comparison, $7B is more than market cap of Lyft, Dolby, and Alaska Airlines. What is happening?
https://stockanalysis.com/list/mid-cap-stocks/
If any of those companies put Ai into their pitch they too would be worth billions
I landscape gardens... With AI.
Pay me.
You've already seen Long Island Blockchain, now get ready for... Alaska AI Lines!
LLM traces are supposedly very valuable. I imagine OpenRouter has one of the most extensive and diverse set of traces in the world.
If your data is leaking to OpenRouter, how many companies would be OK with that? So I doubt OpenRouter are peeking inside the workloads that flow through their proxy. However, they do publish "Top Models by Task" ranking[1] so maybe you have a point.
[1] https://openrouter.ai/rankings#task-spend
The top models by task is self reported. It’s based on the referrer header or some custom one I forget at this moment.
If you don’t include that it just doesn’t count those tokens.
when you use a closed source router you don’t know what it’s doing.
that's why my site TrustedRouter is hosted and end to end encrypted and you can actually see what it’s doing.
openrouter prompt and i/o logging is off by default https://openrouter.ai/docs/guides/privacy/data-collection
the providers you route to will of course have their own policies, which openrouter surfaces to you through the webui and api. you can even configure automatic routing to select providers based on your policy preferences.
Oh that's very interesting. Thanks for the link.
I'd suspect it's related to the rise of good/cheap Chinese models, and OpenRouter is the best way to use them without jumping through a ton of hoops.
Exactly , not even small startup anywhere in us , eu or nearby will be ok to send data to China, but with open router you can create a no China provider selector with one llm call and start using Chinese llms without paying to vendors a single cent .
What’s the angle for stripe , electrify over tokens exchange is the new money flow , and stripe wants to monetize it. 5% tax on any llm token is an amazing deal
> but with open router you can create a no China provider selector
why on earth would you trust no data is sent to China? just because you clicked a toggle?
By that logic, how can you trust OpenAI and Anthropic not to do the same?
For one, openai and anthropic are the top two AI labs and therefore have some reputation at stake. Openrouter is like aliexpress. Any fly-by night vendor can get listed on there.
The Chinese providers also have reputations at stake
Not all reputations are equal.
Good point, the American labs have an awful reputation, and public sentiment is so low it might as well be negative. Plus they've already broken our trust by scraping and stealing everyone's intellectual property to begin with, so they have a lot of catching up to do in terms of reputation and trust.
>Good point, the American labs have an awful reputation ...
Reddit/bluesky users seething at anthropic/openai doesn't matter, because they're not the one making the purchasing decisions for AI. Moreover all the complaints you mentioned applies to chinese AI labs as well, with the extra issue that they're beholden to the CCP. Taking the side of chinese ai labs because they're the competitors of american labs is "enemy of my enemy is my friend" logic, which is just dumb. It's like those people who think the US is bad, so then go and simp for china/russia.
Do you really trust the notorious liar Sam Altman?
Openrouter is very transparent about where your requests are going. Every single one is logged.
You decide the routing if you want.
I trust it because there are at least 7 billion on the line. If it came out that they were violating their contract and sending company secrets to China when they promised they won’t, they would lose literal billions for basically no gain, and possible criminal charges.
Why on earth would you think they are sending data to China after you click that toggle when they have every reason not to?
As a QA engineer with over 30 years of identifying software regressions, I would assume that data would eventually get sent to China even if you click that toggle. One ambitious intern or new hire is all it takes, especially in an age of agent-generated code and 1K+ line PRs.
With that logic you shouldnt even deploy to the cloud at all. Someone in AWS can also accidentally share an S3 bucket with backups to the whole internet, there are plenty of precedents for that
Gemini is also notorious for its hoop jumping, which you can bypass effortlessly with OpenRouter.
Cursor got bought for $60B. This is simply how AI companies are valued.
It's an inflated number due to the AI market, but their selling point is distribution.
Many people and businesses want to experiment with different models, but they don't want to sign up for a dozen different services. Businesses can make it difficult to approve new vendors. If your company is looking at 5 different vendors for tokens and teams can't agree to switch together, OpenRouter comes along with a unified interface and a single billing point.
They also become the point to add value-add services on top in a portable way. They already offer some things like automatic JSON repair, but I can see them adding functionality like leak detection tools, monitoring, alerts, and other patterns that a company can set up once and use with all of the models theirs teams need.
What I'm not so sure about is their moat. They have the brand recognition, but it seems rather easy for someone else to build what they've built. I'm a little confused about why Stripe didn't just build the same thing internally. Acquiring this company gives them an instant boost of 10 million customers for their AI business, which might be key to some financial goal they've got.
>It's an inflated number due to the AI market, but their selling point is distribution.
>Many people and businesses want to experiment with different models, but they don't want to sign up for a dozen different services. Businesses can make it difficult to approve new vendors. If your company is looking at 5 different vendors for tokens and teams can't agree to switch together, OpenRouter comes along with a unified interface and a single billing point.
Isn't that ripe for being picked off by aws or azure? Both already have marketplaces where vendors can offer whatever cloud services they want. Both already offer first party inference service, and have contracts with all the stodgy corps where it's "difficult to approve new vendors". Not to mention they have IAM and SSO built in. Good luck bolting that onto a third party vendor like openrouter.
I can actually restrict how much is spent using a sensible user interface with openrouter. That’s a real pain with aws
You don't want to spend 2 hours in the IAM and Role control panels?
Yeah, but at least currently, openrouter has the latest open source options way faster than AWS does.
AWS Bedrock doesn’t even try to stay on top of new open source models or their API capabilities. We’ve also run into also kinds of frustrating issues rate limiting and slowness issues. GCP is probably the best alternative now that they have actual spending caps, I believe a lot of Openrouter models are run on GCP.
The other popular option is something like LiteLLM, which just has a major vulnerability that left a lot of big corps exposed. After spending a year trying to fight this battle, I’ve decided OpenRouter is worth their cut (and maybe more but don’t tell them).
As a nobody I like OpenRouter because it lets me quickly test out all of the emerging open weight models. I normally use Claude Code but if I run out of usage I'll pull up GLM 5.2 or Deepseek V4 in OpenCode via OpenRouter. Occasionally I'll run Claude's work by a Google/OpenAI model through OpenRouter. If Claude Code becomes too expensive I now know I'll be pretty comfortable with open weight models.
When there is a bidding war for a company, I don’t think fundamentals matter as much anymore..
You can't really compare market caps like that. Addressable market or market share is only one piece of the puzzle, and the other companies you mention are in very different kinds of markets, with very different kinds of products, with very different kinds of costs and margins.
A valuation just reflects what someone thinks about the future cash flows of the business.
But yeah, it does feel a bit crazy; unsurprisingly, AI hype affects valuations of AI companies too. On the other hand, I can see the idea that some people might be betting on the idea that the big US labs are bloated and spend too much money, and that the real money is going to be in serving open-weight models, and/or in automatically combining and routing to different models based on the task at hand.
The market cap of any business operating in the physical realm has to account for enormous liability, especially for one like an airline. Also, their operating costs are huge with much less ability to scale, resulting in much lower potential increase in margins.
Their market share is as big as it gets for that type of AI business.
A good year for Alaska Airlines is 5% sales growth. OpenRouter is growing that much per week I'm guessing.
I prefer using OpenRouter because the money I pay them in premium is far less than the money I lose by not using it on other providers. On OpenRouter, I can switch dollars between models. But for the providers, after evaluating I'm stuck with some number of credits on ones I don't use. This kind of intermediation is actually super useful to me. I must imagine that they can also provide other things that are of value like provider quality measures, ability to swap providers during downtime, etc.
You could just as easily ask yourself how can a middle man for people finding web pages be worth so much, but here we are, and Google is worth something like $4 trillion, mostly on the back of search.
As it turns out, it's very valuable to be the intermediary between a large number of people who want something and a commoditized market of providers. It's the middleman who captures most of the margin.
Now, AI models are not a commodity yet. But things seem like they might be heading that direction. And in a world where they are, Stripe probably wants to be that guy sitting in the middle.
Optionality. This positions Stripe well to compete in AI compute, model development, apps... This is not a simple 1+1=2 merger.
It's pretty valuable to own the customer touch point
Also when people ask questions like this it usually means they're asking questions about a point in time, eg given today's numbers
But valuation should account for trajectory (where will they be in five years?)
In the hypothesized bull case for ai, they benefit dramatically from the secular tailwinds (ie unrelated to their own business strategy) of AI
So one way to consider this is it's a hedge by stripe in case AI becomes as big as some people think it is. And 7b to get in on the ai wave is a good deal in that case. And in all the other cases, it goes to zero and stripe is fine
OpenRouter raised money at a $1.3 billion valuation a few months ago, if the NYTimes reported valuation is accurate.
Going from a $1.3b valuation to a $7b exit in a couple months is an amazing return for those investors. I hope the OpenRouter employees got some decent equity out of this
They should just hold Stripe stock, plus there are always liquidation preferences so it is not so black and white
Historically acquisitions have never really been good for customers. Time for me to look for an OpenRouter alternative? At least they're also as easy to switch from as the model providers they proxy.
Same fear. At least it wasn't Google
Why look for an alternative before you’re affected in any way?
One reason is to avoid a potential boiling frog scenario. I'm already in one with Android, but there are no viable - for me - alternatives (never touching Apple and mobile Linux isn't there yet) so I'm truly stuck. If there's a chance things may start changing at OpenRouter in such a way that months/years from now I'll look back and say "oh, how I wish I'd switched providers", I'd rather switch now just out of principle than be boiled.
what are you looking for in an alternative?
my site is the only end to end encrypted one TrustedRouter.com
Hello ClosedRouter!
Any implication for the privacy and confidentiality of our queries? Openrouter yes sees everything. But as long as they are independent and establishing their core business on privacy and confidentiality reputation, I feel more comfortable with this than with a new broker. AI queries can contain more sensitive information than just financials. And financial informations are protected by law, btw. AI queries, however, are not.
> Earlier this year, Atallah described OpenRouter as the AI equivalent of Stripe.
Not sure I understand how this is strategically aligned for Stripe but certainly an interesting comparison.
OpenRouter uses Stripe to handle payments [1], so this acquisition hopefully will reduce OpenRouter's cost while increasing Stripe's revenue.
[1] https://stripe.com/newsroom/news/openrouter-and-stripe
Stripe seems to appear to be attempting to compete with the US Federal Reserve as the core dollar payment processor while also dipping into the same for tokens. Core value exchange platform, dollars, tokens, whatever. A bit of a less exciting version of Dune. “The transaction volume must flow.”
> Every company in the Forbes AI 50 that monetizes does so on Stripe.
That’s chilling.
Stripe started as a result of the PayPal Mafia.
"In 2011 the company received a $2 million investment, including contributions from Elon Musk, PayPal founder Peter Thiel" [1].
Also fun fact from the Founder's Wikipedia page
"In November 2024, Collison was criticized for visiting Israel and posting an image with the Israeli flag,[31] amid the Gaza war."[2]
[1] https://en.wikipedia.org/wiki/Stripe,_Inc. [2] https://en.wikipedia.org/wiki/Patrick_Collison
Perhaps the US State Dept. used it's strong relationship with Stripe to encourage this purchase as a geopolitical move. They've done so before with companies like eBay and Microsoft to neuter the P2P telephone access network that was "Skype".
Yea they love centralized brokers, exchanges, etc.
Both are in the business of putting a single API key in front of a fragmented ecosystem and charging a convenience fee. This middleman business is naturally coalescing.
The nature of the ecosystem also means that pricing is closely tied to "procurement" which could be routing, limits, whatever at a company level.
If stripe wants to be _the_ one that charges that fee, they either have to continuously try to ensure that all the different middlenen use stripe (most of them do!) but even better is to acquire the largest middleman.
You don't want someone else who happens to do all the routing+limits+policies, end up not using stripe. They already have this hold in existing stripe financial products where they apply all the policies, and everything goes through them.
It is also an easy deal from an investor point of view.
As a payments processor, Stripe has conversion data. They can route based on ROI with end-to-end A/B testing or online optimization. There's a tricky credit assignment / multi-touch attribution problem to solve for some use cases, too, that could benefit from transfer learning across their customers. Also, the (unstructured) data they collect from OpenRouter users can be used to improve many other dimensions of their business, including fraud detection. Lastly, both businesses look fairly similar in unit economics if you squint hard enough; for example, they both collect a ~3-5% transaction fee from most customers.
As opposed to PayPal which is now talking in front of the mirror.
https://finance.yahoo.com/markets/stocks/articles/breaking-p...
The strategy here is that the same services that Stripe offers for payments have corollaries in the LLM world. Security, user management, perhaps injection attack monitoring, etc.
However I think they dearly overpaid for this as the core technology behind Stripe (fraud detection and integration with global banks) is hard to replicate. Even with the features mentioned above I think the technologies behind OpenRouter are vastly easier to replicate, perhaps even trivial now.
This kind of makes a lot of sense. While my first stake was surprise that a proxy is valued at such a high rate, But it really comes down to the terms if open router is allowed by their terms to see the prompts and responses, Then that's very, very valuable data today for any model improvements and other opportunities that people might be looking for.
I'm surprised OpenRouter went for 7B, while fal.ai just raised at $8b, despite fal having far less traffic (semrush reports ~5x more traffic to openrouter). They seem like very parallel businesses, just with focuses on different models (creative models vs LLMs).
LLMs have zero margin.
Media models have crazy margin.
Have you seen FAL's pricing? It's absurdly marked up. 50% or more.
fal actually serves models while the router is just a side project. last ARR 400m vs openrouter 100m
If someone from Stripe / OpenRouter is listening, please let me provide Open Router OAuth to my users and let me take a cut of their usage. Apple takes 30, I would take whatever the users are willing to pay. 5 to 15 is reasonable.
I'm still amazed at Apple's 30%, it's just so brazenly high (like: "fuck you, we're Apple, we do what we want and you'll lick it up"). Unfortunately, too, other companies follow Apple's lead, and so this massive rent seeking percentage has become normalised.
I can totally agree that Apple sucks for doing this, but you seem to insinuate they started this. As always Apple wasn’t first not even tenth: Nintendo, Steam, Xbox, Kindle all did it prior to “the platform economy” 3)
Didn't know there was prior art of that 30% scale, as I generally avoid as much of that shit as I can, Apple is the one I hear about the most.
I amend my prior comments (with actually being able to amend them) to Apple sucks as much as the rest of the rent seekers, and in reference to Aeolun's comment, it doubly sucks that enough people choose to lick it up that the rent seekers don't feel ashamed of their rent seekery, but in fact profit handsomely from it, thus encouraging others to follow suit.
It feels ridiculous that you can't buy directly from the software publisher - like the old internet when they were called programs. I feel I may be on the end of pointing and laughing at the old man, but the situation now is literally a gatekeeping middle-man that gets a 30% cut. WTF? Nice work if you can get it.
Makes me feel like a rube (in a significant majority, mind) for the few games I've bought off Steam. Eesh, even GOG is 30%. Epic's seems reasonable at 12% and only after the first million in revenue per year.
Software publishers need a 'donate' button on their websites where erstwhile pirates can cough launder cough their procurements.
I mean, the problem isn’t the rent seeking. It’s the licking it up part.
For the first part they have https://openrouter.ai/docs/guides/overview/auth/oauth (which I haven't used). That gives you an openrouter key for that user (openrouter bills the user).
What value are you providing that you'd take some further cut? You could take the money from the user via stripe and then pay your own openrouter bill instead.
we added that to our router https://trustedrouter.com/sign-in-with-trustedrouter
I’ve stopped using OpenRouter this year. And I stopped using different models for testing. Anthropic won. I’ve even let my OpenAI API credits expire (which is a ridiculous policy in itself) because I had no need for them.
Congrats to the OpenRouter team.
I vividly remember back in 2023/2024, I was sitting on my sofa thinking, "there has to be a market to abstract data access layer, something like Terraform but for LLM provider to avoid vendor lockin". Only if I had access to capital to build something like that at that time, I could have been my try for an AI start up.
Congrats to all parties. Makes me more likely to use openrouter and excited for where stripe is heading.
Does this mean they will be censoring openrouter?
Is there an open source alternative for when the censoring begins?
I don't really understand the value -add, so far "smart" routing currently produces worse results at higher cost and static routing can be done yourself trivially. Plus the risk of sending the most sensitive data you have to a third party.
I don't think you understand what Openrouter's product is.
Even ignoring routing, it's still very 'nice' to be able to buy almost any model in one place.
Amazing for enterprises, or companies who are letting users choose from a variety of models for ai tools.
source? most benchmarks show "smart" routing producing near-identical accuracy and 50-75% cheaper costs.
It was a nice service. A bit worried (probably irrationally) how it's gonna be in the future.
Congrats to the team anyway.
Ramp LLM Router Stripe LLM Router(?)
What’s up with finance companies getting into LLM routers?
They have the unique advantage of seeing where money is flowing and are positioning for the most likely future based on their analysis.
Everyone seems to be asking why this a great strategic fit for Stripe.
Stripe can now provide tools to every product that sells metered AI usage and take a cut. This isn't about extracting a small percent on the tokens flowing from your coding agents to your model of choice. It's about all the products that are going to come to market and monetize metered usage.
Want to analyze your 2026 tax return? Use VisorAI's Tax Agent and pay only for what you use. Stripe provides all the accounting, payment processing, ships money to vendors, and takes a bit off the top. Trillions of dollars moving from the labor market to the token market? It's $10B per point in fees for every trillion.
sorry did i just read a tax agent ad in a HN comment
The reason OpenRouter can build value even if there are 100 clones is switching costs and flexibility. AWS Bedrock may be great for large enterprises, but flexible startups will use OpenRouter for ease, and now they have Stripe's distribution. Once you use OpenRouter, you won't switch because you become embedded in the logs and cost-saving systems.
AWS Bedrock - Microsoft Teams
OpenRouter - Slack
https://s-1.vercel.app/posts/why-openrouter-can-be-the-next-...
A large missing link for AWS Bedrock is Gemini model support [0]. And this is a vital feature, because Gemini itself (as far as I'm aware) does not have a real-time daily or weekly hard spend cap in the case of misconfigured or compromised callers [1][2].
Meta's closed source models, too, are only available on OpenRouter (and it's touted as a first-class recommendation on [3]).
So, in many ways, OpenRouter is the only game in town to access some of the highest-funded model families on the market through a single API endpoint.
I'm unsure why AWS is ceding this territory (there have to be ways to allow users to opt into the lesser governance inherent in using third party providers, while not diminishing the brand's governance requirements). But, since they are, it's a meaningful opportunity left open.
===
[0] https://docs.aws.amazon.com/bedrock/latest/userguide/model-c...
[1] https://www.reddit.com/r/googlecloud/comments/1szv8cr/warnin...
[2] https://www.theregister.com/ai-ml/2026/05/13/google-users-fi...
[3] https://developer.meta.com/ai/products/meta-model-api/
what are they? all you have to do is change a URL and a token in your agent harness.
Well, you have the logs and specific cost control measures and setups in place. You would lose all that in the event you attempted to switch, plus Stripe owning OpenRouter gives them stronger distribution and further solidifies the switching costs in the broader integrated solution Stripe offers - https://s-1.vercel.app/posts/what-stripe-can-become-broader-...
As an openrouter user this is better than most alternatives I guess
yeah, i bet this seems like a good deal in a few years, we'll see. remember Poe haha that was a similar kinda thing but openrouter waited until the market was mature
RIP free deepseek access.
Free model usage is controlled by the LLM provider, not OpenRouter. Both parties have their own incentives to allow it and that would not change with different owners.
The free DeepSeek depended on full data sharing straight to China.
Terrible news for the market. Even as someone who dislikes AI, it's terrible. Stripe is a horrible company and more market consolidation is very worrisome.
Hope someone clones it and they take a loss at it.
Also someone start counting days until they implement mandatory KYC and AML into request processing.
OpenRouter should first fix their support.
No support exists when things go wrong!
They do reach out pretty quick when your usage takes a huge dive after you move to one of the single provider plans that offer hard to believe possibly unsustainable value.
I had consistent problems (5 months back) with OpenRouter and other providers. And very hard to debug. I used the provider directly (Google) and worked perfectly.
i feel really happy for the founders! congratssssss
I hate people linking to paywalled articles, wish it was against the HN terms.
Here’s another link to an article about the acquisition; https://siliconangle.com/2026/08/16/stripe-reportedly-finali...
This is even before them acquiring PayPal. If Stripe bought PayPal in 2022, it would be immediately blocked. Not this time.
One of the only near monopolies that is seemingly allowed and it is even praised.
But who cares. Nothing to see here (as long as AGI is coming it doesn't matter anyway).
Why would PayPal acquiring OpenRouter make it any more of a monopoly than it is? Not saying it is or isn't, but OpenRouter isn't even a financial company.
I am saying it is a near monopoly as Stripe is the one that is close to acquiring PayPal in the first place. [0]
[0] https://www.wsj.com/business/deals/stripe-advent-in-talks-to...
Le sigh.
The enshitification will be inevitable. Soon this wonderful resource will no longer exist.
Le sigh.
The enshitification will be inevitable.
Back to my own brain for my coding projects.
We buying inflation burritos with tokens ain't we
I still find it hilarious that AI is so bad you need something to sit in front of it to pick models for you. And that's a normal, accepted thing.
That's not a very common pattern, to be honest. A bunch of people have been experimenting with automatic model routing recently but mainly as a cost optimization, since tokens for the best models have got expensive once you start piping millions of tokens through them.
I haven't seen much evidence that model routing is being widely used yet. I think it's still more of an experimental mechanism right now.
OpenRouter is a proxy, not an automatic router. Rather than building API clients for five different AI providers, you build one client to OpenRouter, and switching models become extremely easy. This matters when new models are coming out virtually every day.
Does that means if by any chance all frontier model companies agrees to use an open protocol for the API open router would go out of business?
All the model companies except kind of Anthropic (and even they half-assedly do) implement the OpenAI API. It's not an open standard but, like the S3 API, it effectively is.
And, to answer your question, no. The existence of a common API makes it trivial to change zero code and send requests to a different model.
The Anthropic messages API is a competing standard (it's just better than the OpenAI API which even OpenAI has moved away from) and some Chinese providers use it as their standard.
Isn't this what LiteLLM is doing? And is Open Source? Maybe I am asking a dumb question because this is the age-old SaaS vs OSS debate, but I am struggling to find the angle here.
no. i don't know if openrouter even guarantees consistent protocl across models.
the value of openrouter is it offers centralized billing. you can route your calls to any provider you want, test a whole bunch of models against each other, and you just get one bill from openrouter. switching to a new model, or a new provider of the same model, doesn't mean setting up a new billing account with a new provider.
I see, that make sense, thanks for the explanation. Because otherwise I was thinking their whole business model would be just to provide an API tool library.
It's more complicated than that. Lots of providers use an "OpenAI-ish" API, but many of them have subtle differences in things like tool calling or thinking blocks. OpenRouter normalizes the wire format.
OpenRouter does more than just proxying; they also aggregate providers for open-weight models, which has a stabilizing effect on pricing and gives you protection against a single provider's downtime.
Dumb question but aren’t they all OpenAI API compliant? I thought that was pretty standard, eg I know Anthropic works with it. Also Claude Code can work with different models, they’re probably not using OpenRouter for that?
Effectively but not quite. Anthropic does mostly support the OpenAI API but there are minor feature gaps. You have to use Anthropic's bespoke API to use everything (which may or may not be important to you).
There are some gotchas with it too. e.g. The "system" role message isn't supported in the Anthropic API. It's not the biggest deal but it's a potential footgun.
On the other hand, the openrouter SDK + API takes care of this for you: Their one gateway has ways of enabling provider-specific features. You can switch the model name and pretty much be good to go.
Claude Code does not work with other models out of the box. You need a wrapper around Claude Code that translates other model requests + respondes to what the harness expects.
That makes sense, though I wonder if Claude is a bit of an outlier there because they have the developer audience?
As an example, GLM document being committed to compatibility with OpenAI: https://docs.z.ai/guides/develop/openai/python
My uninformed guess would be that everyone except the frontier labs is probably motivated to do that too.
Similarly they document configuring Claude Code for their models, so I assume they implement Anthropic’s interface there too: https://docs.z.ai/devpack/tool/claude#manual-configuration
The APIs are similar but not the same, which is to say: different. Definitely not as simple as swapping out the base URL.
Claude Code has no built-in way to use non-Anthropic models. It does support overriding the API URL with an environment variable, but there are drawbacks. OpenRouter is not used within Claude Code at all.
I don’t think it “being bad” is the motivation. It’s about saving money and pooling cheaper resources.
I think that OpenRouter's goal is to have the user pick the model, so there's manual configuration for the user control and cost benefit. Maybe you meant OpenCode Zen? But let's be honest, "AI" is definitely incredible, not "hilarious", or "so bad".
No, you don’t. But most have fomo and use them.
>I still find it hilarious that AI is so bad you need something to sit in front of it to pick models for you. And that's a normal, accepted thing
My lord. Of all the terrible, uniformed takes the HN posters are spewing, this is the worst one I've seen in a while.
The anti-AI crowd really are clueless, eh?
I'll save you $7b.
https://api.anthropic.com https://api.openai.com/v1
They don't accept bitcoins
Neither does OpenRouter with Stripe for what it's worth.
What do you mean "OpenRouter with Stripe" are you suggesting they will remove the Bitcoin/Crypto payment rail?
I mean that OpenRouter uses Coinbase to handle crypto for them.
currently my router takes stablecoin through Stripe
TrustedRouter.com
where are the cheaper chinese models?