# He Built the World's #1 Open-Source Coding Agent Y Combinator · 2026-07-24 Jay V, founder and CEO of Opencode, explains how his open-source coding agent grew from 650,000 to 13 million monthly active users in 2025—a 20x increase—while processing 7 trillion tokens per day and reaching $40 million annualized revenue. Anthropic's attempt to block Opencode users from using Claude Code subscriptions inadvertently fueled the growth by equating the two products in users' minds. Opencode's success hinges on offering choice across 70+ models, including open-source alternatives like DeepSeek and GLM, which are cheaper and faster than frontier models, making coding agents accessible globally—especially in developing countries where a $200/month Claude Code subscription is prohibitively expensive. Jay shares that 12 Fortune 500 companies use Opencode, often discovered via bottom-up adoption. The product's terminal-first design and open positioning were deliberate choices. Jay also recounts his 16-year entrepreneurial journey: the same legal entity (incorporated 2010) applied to Y Combinator nine times before acceptance in 2021, and each prior failure—from a serverless platform to a terminal coffee shop—contributed to the product and market instincts that made… ## Questions this episode answers ### How did Anthropic's attempt to block OpenCode actually fuel its growth? In January 2025, Anthropic tried to block OpenCode by rejecting Claude Code API requests when the system prompt contained the word 'OpenCode.' Jay V explains this inadvertently put OpenCode on the same pedestal as Claude Code, attracting massive attention. The controversy helped drive OpenCode’s monthly active users from around 650,000 at the year’s start to 13 million by June. [3:53](https://ycombinator.podhood.com/eaaf97ed-46fd-4d8d-94e8-28402cc5e156?t=233000) ### What is the story behind OpenCode's founder's 16-year journey to success? Jay V and his co-founder Frank started their company in 2006, incorporating in 2010. They applied to Y Combinator nine times across a decade before being accepted in 2021 with a serverless platform. The same legal entity persisted through multiple products. Jay attributes their persistence to stubbornness and gradual learning, even living with parents when funds were low, before OpenCode’s explosive 2025 breakout. [34:18](https://ycombinator.podhood.com/eaaf97ed-46fd-4d8d-94e8-28402cc5e156?t=2058000) ### Why is OpenCode especially popular in countries like China, Brazil, and Indonesia? Jay V notes OpenCode’s $10/month subscription makes coding agents affordable in developing countries where a $200/month Claude Code subscription is very expensive. Additionally, OpenCode supports models from local labs, such as Chinese models, appealing to home-country users. This drove adoption, with China at 17% of traffic, Brazil 5%, and Indonesia 4%. [13:29](https://ycombinator.podhood.com/eaaf97ed-46fd-4d8d-94e8-28402cc5e156?t=809000) ## Key moments - **[0:00] Intro** - [0:01] "Most people in the world still haven't experienced the magic of a coding agent," says Jay V. - **[1:24] Growth** - [1:32] OpenCode hit 13M monthly active users in June, a 20x increase since January 2025, and processes 7T tokens daily. - [2:19] OpenCode's inference revenue hit a $40M annualized run rate, up from $31M earlier in 2025. - [2:41] OpenCode's subscription product launched in March 2025 reached 160,000 monthly subscribers, contributing $18M ARR. - **[3:42] Blocked** - [3:53] Anthropic's attempt to block OpenCode in January 2025 inadvertently equated the two products and drove a surge of new users. - **[5:30] Vision** - [8:09] Kimi 2.5 became the first open-source model to surpass Anthropic in OpenCode usage in Feb 2025, showing they were ready for real work. - **[8:55] Data** - **[13:29] Global** - [13:38] China accounts for 17% of OpenCode's users, making it likely the only YC company with significant Chinese usage. - [16:02] A dozen top Fortune 500 companies use OpenCode for model choice and flexibility, despite unlimited token budgets. - **[16:33] Economics** - [19:25] Enterprises proactively ask OpenCode to complete security questionnaires because their employees are already avid users. - [21:35] Ramp built a Slack bot powered by OpenCode's embeddable agent server, impressing the OpenCode team who hadn't built that internally. - [23:32] OpenCode uses token subsidies for its free tier as customer acquisition cost, like frontier labs but with better unit economics. - [27:39] DeepSeek's success comes from specializing in low cost on the quality-cost-performance spectrum, says Jay V. - [28:17] OpenCode is the largest customer for most open-source model labs by token volume. - [28:59] OpenCode's global user base creates a stable 24-hour GPU utilization cycle, improving unit economics over competitors. - **[29:55] Design** - [30:58] OpenCode built models.dev, an open-source database of 70+ AI models and providers, to enable instant support for any model. - [33:16] Jay V previously built terminal.shop, a complete storefront for ordering coffee over SSH, showcasing terminal UI expertise. - [34:18] OpenCode is a 16-year-old legal entity that applied to YC nine times before achieving breakout success. - **[34:21] Journey** - [34:41] Jay V's first YC interview was in the batch after Airbnb, with PG conducting interviews and Airbnb founders in the waiting room. - [38:46] OpenCode's building-in-public philosophy, inspired by YC's Dalton Caldwell, makes their journey feel like a reality TV show to the community. - [41:15] "Partly maybe being a little stubborn," says Jay V on why he never gave up after 16 years. - **[42:19] Try** - [42:19] YC partner Harj Taggar, a diehard Claude Code user, switched to OpenCode and was impressed by its performance with open-source models. - [43:13] Q: How should developers try OpenCode? A: When a new open-source model launches, OpenCode lets you pick it from the model picker instantly. ## Speakers - **Diana Hu** (guest) - **Harj Taggar** (guest) - **Jared Friedman** (guest) - **Jay V** (guest) ## Topics AI & Machine Learning ## Mentioned Anthropic (company), Ramp (company), Y Combinator (company), Claude Code (product), Codex (product), Cursor (product), DeepSeek (product), GLM (product), Kimi (product), Minimax (product), OpenCode (product), OpenRouter (product), SST (product), models.dev (product) ## Transcript ### Intro **Jay V** [0:01] Most people in the world still haven't experienced the magic of a coding agent. **Harj Taggar** [0:05] This is just an unprecedented market, like the market for intelligence has not existed before. Everybody should be thinking in a positive sum, grow the pie mentality. **Jared Friedman** [0:12] It's huge in developing countries like Indonesia is 4% of your traffic, Brazil is 5% of your traffic, places like Vietnam, you know, like places where a $200 a month Claude Code subscription is like very expensive. **Harj Taggar** [0:27] You really know you have product-market fit when like enterprises are bugging you to sign the security agreement so they can use your product. **Jay V** [0:33] Just be like, please do this, because I don't know who you guys are, but a bunch of us are using this. We want, you know, everybody in the world to have that aha moment with a coding agent. **Harj Taggar** [0:51] Welcome back to another episode of The Lightcone. Gary's out traveling today and will be back next episode. Our guest today is Jay V, founder and CEO of Opencode, an open-source alternative to Claude Code that works with any model you want. Opencode has been growing at an astounding rate this year. They're now at 4.6 million weekly active users, which is actually pretty close to Codex's numbers. Today we're going to talk about what's driving their wild growth, and also Jay's winding road to get here since the company went through YC back in 2021. Jay, thanks so much for being here. **Jay V** [1:22] Thank you. Thank you for having me. **Harj Taggar** [1:24] Why don't we just start with kind of like the crazy scale you guys are at? Maybe tell us about any stats you can share with us. ### Growth **Jay V** [1:32] Yeah, yeah. I mean, you mentioned the weekly actives. Our monthly actives, I think we ended June with around 13 million or so, and that's around a 20x increase since the beginning of the year. We also recently started processing around 7 trillion tokens per day. For context, OpenRouter does a total of around 6 trillion. I think again, sort of beginning of the year, we're probably at around 300 billion or so. And then in terms of sort of our revenue off of our subscription product, and if you were to pay per token with our inference, and you took, let's say, June's data and extrapolated for the year, that would be around 31 to 33 million or so. And if you took last week's data and extrapolated for the year, that's around 38 to almost 40 million. And that's the inference part of our business, so the way we sort of make money. We had launched that, call it end of September, early October last year, so about eight months to getting to around 40 million or so. I guess to sort of round out the numbers, our subscribers, so these are people that pay for a monthly subscription with OpenCode. We launched that product early March, I think end of February, and that's grown to around 160,000 monthly subscribers. Yeah, and around that, that accounts for about 18 million of sort of the annualized revenue. **Harj Taggar** [3:07] I saw a tweet from, I think he is Tibo, I think he's a lead engineer, at least one of the main engineers on Codex, saying that something like, I guess, 5% of all Codex subscribers choose OpenCode as like the main harness to actually use the underlying API. **Jay V** [3:25] Yeah, yeah. So Codex officially, you know, supported OpenCode. What that basically means is that you can use Codex's subscription in OpenCode, and yeah, and a bunch of their users use OpenCode directly to take advantage of their subscription plan. **Harj Taggar** [3:42] And I think they did thatright after kind of some of the back and forth you had maybe with Anthropic and Claude Code. Like, tell us about what happened and how that seemed like it really fueled your growth. It was an inflection point for you guys. ### Blocked **Jay V** [3:53] Again, to sort of contextualize this, we were at around 650,000 monthly active users to begin the year. And the first week of January, we started to sort of hear some rumblings around Anthropic trying to clamp down on people using OpenCode but using Claude Code subscriptions on there. And this was a very sort of common way to kind of use the Claude Code subscription. And the way I think they sort of blocked it, or they were trying to block it, was if the system prompt mentioned literally the word OpenCode, they would sort of reject the request. You know, from our perspective, that sort of makes sense. You know, they're sort of subsidizing usage. That's kind of what they want to do. But of course, a lot of users weren't happy. And when they had kind of done that, what it inadvertently did was it put OpenCode and Claude Code on the same sort of pedestal. It like equated the two products in some ways. And even for the people that weren't using OpenCode at the time, you know, they sort of took notice of the fact that, you know, Claude Code is taking that kind of an action against OpenCode. **Harj Taggar** [5:01] You got new users because people heard about you for the first time and they're like, oh, Claude Code's like banning this thing or like clamping down on this thing. **Jay V** [5:08] Yeah, or that it is like worth looking into, that, you know, it's not just one of the other dozen or so coding agents out there. Yeah. **Jared Friedman** [5:15] It's funny how often this happens in startups. **Harj Taggar** [5:17] Same thing happened with Instacart when Amazon bought Whole Foods. It was like, oh, this is like the death of Instacart. And the death of Instacart became like this meme. But the meme actually just like drove all the grocers to check out what Instacart was. And then they just went through this like explosive growth of signing up every grocer in America. So it seems like actually a lot of your growth is global, sort of across the world. Tell us a bit about that. ### Vision **Jay V** [5:41] Yeah, so the premise of the product is that, you know, most people in the world still haven't experienced the magic of a coding agent. And, you know, it's been almost a year, and I'm sure, you know, it's sort of hard to remember maybe for you guys as well. But the first time you had this experience, it's a magical experience. And when we had felt that for ourselves, we sort of recognized that, you know, how important that moment is in sort of tech history, I guess. It comes along once every sort of generation or so. And the way we sort of looked at that was, let's take that, let's take that to as many people in the world as possible, you know, have them sort of experience something similar. Because the frontier models and the frontier labs charge so much per token that it's going to be hard for a lot of people across the globe to have that experience. And we wanted to make sure they had that with us in some ways. **Harj Taggar** [6:37] I guess there's a point where, is it true like early on there was sort of like seemed like there was a big gap between the open-source models and the frontier models? Is that true kind of when you were first launching the product? **Jay V** [6:49] Yeah, for sure. I think when we had first launched, it was mostly, hey, you're using your Claude Code subscription with Claude Code, come try that with OpenCode. And then, and so this was back in June of last year. And by about August, September or so, we started to see the first crop of open-source models. And you had this sense that like, oh, they're maybe six months behind, you know? And of course, there's always been sort of a gap between them, but that was sort of the first point when people were like, there's the GLMs of the world, the Kimis of the world, the Minimaxes of the world, and it seems like there is now an alternative. And as that started to happen, that would trigger a wave of users coming in trying out OpenCode because that was probably the only way you could try out some of these multiple models. And as that gap started to shrink or the fact that the open weight models became good enough for real work is when it sort of became viable to use OpenCode with them. **Harj Taggar** [7:47] Yeah, I think I remember you maybe a few months back, you were saying that, I'm going to get thisright, that even though the open-source models are obviously cheaper to use through OpenCode, you still saw more usage from the leading frontier model, except was it when Kimi 2.5 came out? **Jay V** [8:05] Yeah. **Harj Taggar** [8:06] That was the first time that it was equivalent or it flipped. **Jay V** [8:09] Yeah, that's exactly it. I think it was like probably 2.4. I forget the sort of exact model, but yeah, this was February of this year, where for a four-week span, and this had never happened in our data before. And I think we're sort of fortunate to be able to see this global usage. And so a lot of times we see the comparison of, you know, these open-source models versus some of the frontier ones. And we had noticed for the first time that a bunch of users were just using Kimi way more than they were using the Anthropic models. It was sort of Sonnet plus Opus at the time together. And that's the point when we were like, oh, I think we should launch a subscription product because now maybe these models actually do make sense for real work. **Harj Taggar** [8:55] And so speaking of that, like you guys have this incredibly unique insight, like probably the only, you're the most, you have the best data on just how these models are being used by engineers across the whole world. And you released a bunch of this data. ### Data **Harj Taggar** [9:06] So maybe we could just like look through some of it and pluck out some more interesting insights. **Jay V** [9:10] Yeah. So this is, you know, you can go to opencode.ai slash data. I think we started to publish this about a month ago or so. This is basically taking all of the usage on OpenCode Go, which is our subscription plan, where you pay $10 to be able to use any of these open-source models. And this breakdown specifically is by token volume per day across the different models. So what we sort of see here is that DeepSeek Flash is the one that is used a lot. And there's some sort of details here I can kind of go into why that is sort of the case. But if you just sort of look at the top three, we're seeing sort of the two DeepSeeks plus GLM. And you can kind of see all the hype that GLM has been getting sort of lately being sort of reflected here in these charts. **Diana Hu** [9:57] Which the data says otherwise as opposed to all the Twitter chatter about the GLM being taking over DeepSeek, this is telling a different story. **Jay V** [10:07] Yeah, yeah, it is. I can show you a different breakdown here. So this is by unique users. And this was the thing that Herd you were sort of alluding to. We get to see actual usage data for each of the users as opposed to with maybe like an OpenRouter where you're seeing it aggregated across a bunch of services or other products even that are sort of internally using it. In this case, it's, you know, each one of these is sort of an actual user. But yeah, what's sort of fascinating is if we look at the market share graph, so this is breaking down for each of these labs, the amount of token volume they're doing per day, you know, comparing that sort of across them. You can kind of see the DeepSeek one dip around the time GLM sort of comes out, but it seems to sort of bounce back up after. And we've sort of some theories around why that's sort of the case, but that's sort of an interesting fact. And then if I go back up to the users one, this is also kind of fascinating in that it might be a little bit hard to see here, but if you look at the sort of top three unique users per model, you've got Flash at sort of 38K, and then there is DeepSeek Pro at 31K and GLM 5.2 at close to, you know, 30K as well. And that's interesting because GLM is sort of on par with one of the DeepSeek models, but the fact that there are two of them makes it a little bit different. I'll sort of caveat this by saying that DeepSeek Flash, being as cheap as it is, allows a lot of users to extend how much they can use a coding agent. Because as they get closer to, let's say, their daily or weekly limits, they can switch over to one of these very cheap models, in this case, DeepSeek Flash, to sort of get the rest of their work done. Again, this is very different from the way we sort of think about coding agents and LLMs here in the Valley and SF in the West in general. **Harj Taggar** [12:13] What are you seeing? I mean, we definitely are the other extreme end of where it's sort of like. **Diana Hu** [12:19] Token maxing. **Harj Taggar** [12:21] But for the users you see, which to be fair, it does seem like there's a general vibe shift, especially in the enterprise world towards like more token budgeting. What do you see? Is it as simple as people once they approach their usage limits, they switch over to one of these models, or is there more going on? **Jay V** [12:38] Yeah, there's a few different things. I think people do try and optimize for things. One of the reasons why originally Kimi had sort of taken off was it was being hosted in a way that the tokens per second was a lot higher than what you would get out of even an Opus. And so it was just a drastically faster model. It felt like you were working almost in real time. And some of these models tend to exhibit those qualities, which makes it, you know, characteristically a little bit different from using some of the frontier ones, for example. So there's a little bit of that, but yeah, cost is obviously a big driver. And then the other one that pops up every once in a while is, you know, people sort of get a feel that GLM 5.2, for example, is better at front-end design as compared to some of the other models. And so that ends up sort of driving some usage as well. **Harj Taggar** [13:29] You also have a really interesting breakdown by geo to show like where the users are. Yeah, so who are your users? Where are they coming from? ### Global **Jay V** [13:38] Yeah, so again, some sort of context here is that, you know, we had launched the OpenCode Go plan to be able to serve the sort of global audience. And you can kind of see that with sort of China being kind of number one at 17%. **Harj Taggar** [13:54] That's crazy. I feel like you're probably the only YC company in history that has meaningful usage in China. **Jay V** [14:01] It's also fascinating because a lot of these models are Chinese. So, you know, in their situation, they're trying to use the ones that are being built in their country. And OpenCode kind of gives them the choice to be able to do that. So that's sort of interesting as well. I think the US one is actually interesting to us because when we had built this plan originally, we weren't thinking about the US. We weren't thinking about the states because, you know, as we're sort of saying, you know, people here just throw money at it. But then, no, it turns out it's growing really, really well in the states as well. And maybe, yeah, that speaks to the sort of vibe shift of like maybe you should be a little bit more conscious with the tokens. But then there's the other side of this where people want to use some of these. So when GLM 5.2 was getting popular, a lot of people are trying out our subscription plan because it's one of the options to sort of do that. **Harj Taggar** [14:49] So because it's so much cheaper, it's huge in developing countries like Indonesia, it's 4% of your traffic, Brazil is 5% of your traffic, places like Vietnam, you know, like places where a $200 a month Claude Code subscription is like very expensive. So that makes sense that, you know, sure. But what you were telling me earlier was that in addition to that, there's a lot of like large US companies that have basically unlimited budgets for tokens that are also using OpenCode. Can you talk about that and why, like who's using you and how come those people are using you? **Jay V** [15:24] Yeah, this is what we had seen early on before some of these open-source models even sort of took off is that a lot of companies would start using OpenCode because they didn't want to be locked into using a specific model or a specific harness. So, you know, some users just wanted more choice. And this was a good neutral option for them that allowed them the flexibility in the future to switch to whatever they sort of wanted to. **Diana Hu** [15:51] And I think you had a crazy stat. It was something like a dozen of the top very forward Fortune 500 companies are using you and have a significant footprint. **Jay V** [16:02] Yeah, it's funny. We get DMs, we get emails sort of internally about, hey, you know, XYZ company here, we've got a few thousand people using OpenCode now. Please don't share this publicly. But I think most of it was sort of going on is like, yeah, there's a bunch of choice there. And we can probably talk about this later, but we're very intentional with our product design. We want something that everybody uses every single day and we hold that bar fairly high. And in those instances, I think that's what's sort of resonating with people as opposed to the cheaper tokens thing. ### Economics **Diana Hu** [16:33] I'm very curious on a slightly different topic. There's been a shift for all these companies like you in terms of the AI token economics. Because in the old world of companies, other B2C or B2B companies, there was a lot of a CAC that used to be based on ads. And now the equation is different, it's based on token to acquire users. But also even more weird, the system that you're describing where there's some whales that pay for a lot of it at some point. And there might be a lot of churn, but it doesn't matter because as long as you have the power users and experts really using it, they convert these large orgs. Yeah. And I think that big labs can subsidize that,right? Which is what effectively Claude Code and Codex have done. They can subsidize it, but you have a magic formula to skip all this. **Jay V** [17:29] The broader context here is that for you to use AI and especially these coding agents because of the amount of tokens they use, to use them well, you have to really sort of understand them. And this is the experts thing you're sort of talking about. And to get there is fairly expensive from a per-token perspective. **Diana Hu** [17:48] Token maxing is expensive. **Jay V** [17:49] Right. It's very expensive. And that is a chasm that is very hard for a lot of people and a lot of companies to cross. And so what, you know, the Anthropics and the OpenAIs of the world do is they subsidize it so that people are able to cross that. And then, yes, like the whales idea here, a certain percentage of them cross it, get to that point, are spending the crazy amounts that you sort of see. And then it makes sense. The entire sort of funnel then sort of makes sense. But when we had sort of approached this, we had thought about it from a product perspective where it was, you know, we sort of talked about, we want, you know, everybody in the world to have that aha moment with a coding agent. So that's sort of our free tier. And then because the open-source models were now cheap enough and good enough for real work, you want a subscription plan that allows them to do real work with it. And that is sort of the thing that allows somebody to buy into, okay, now I can justify spending so much more to potentially redo some of the processes within my company to take advantage of these coding agents. And that's when hopefully, you know, some of these sort of turn into whales. **Harj Taggar** [19:01] And is that, sorry, and that's what you're seeing? Like you're seeing people come in through the free tier to try it out and then become advocates to be like, oh yeah, like we should adopt this at our like Fortune 500 company. **Jay V** [19:10] Yeah, I mean, it's to the point that we'll, you know, in sort of the older sort of SaaS enterprise world, you would have this procurement process that a lot of them sort of go through where, you know, somebody sort of reach out and you have this whole dance that they do. In our case, the inbounds that we get are typically just like, hey, there are a bunch of people at this company, at our company using you guys. Can you fill out the security questionnaire? And we're just like, oh, first off, I didn't know. I didn't know we had users there. But secondarily, give us a second. **Harj Taggar** [19:44] You really know you have product-market fit when like enterprises are bugging you to sign the security agreement so they can use your product. **Jay V** [19:51] Just like, please do this because I don't know who you guys are, but a bunch of us are using this. And so it is a little bit different now, yeah. **Harj Taggar** [20:00] Once you're through the procurement and admin side of things, what are the enterprises asking you for? Are they asking you to, are they trying to pull the product in a different direction? Because that often tends to be something that open-source companies have to think through and be careful about. **Jay V** [20:12] What's interesting here, I think, is these coding agents are at the core of how an LLM does work. And so a lot of times when we get these enterprises sort of reaching out to us, it's because they're trying to figure out where else they can use it. There's the obvious one, yes. We've got a bunch of developers at our company that are using OpenCode. Just figure out how we can officially use it. And then on the flip side, it is, oh, there's some non-technical people that would want to be able to use this as well. And then, you know, secondarily, there's this, our product is probably going to be using a coding agent as a part of its core loop. Can we sort of use it there? So yeah, so we get some sort of pull there. The other bit of pull that we see is mostly around just managing tokens a little bit, but being like, hey, there's certain organizations within our company that don't need the frontier models. Can we sort of limit access there or have some more creative ways of managing token spend? And we see pull there. And so, you know, there's some sort of questions around that. The strange one we had gotten recently was somebody wanted to be able to, you know, mostly just have really good visibility of exactly what everybody is doing at the company. And then that sort of gets into questions of like, you know, is that something we want to build? **Harj Taggar** [21:33] Didn't Ramp use you in sort of an interesting way? **Jay V** [21:35] Yeah, I think Ramp was very sort of forward-thinking. They had published a blog post, I think this was in December of last year, but they had reached out prior to that where a team within Ramp had kind of built this Slack bot that was essentially running OpenCode behind the scenes. And it was just sort of incredible to see. It was mind-blowing, partly because we hadn't even done that yet internally. And they were showing off a use case that was definitely pointing towards the future. **Harj Taggar** [22:03] What does that mean exactly? I think it can be hard for people to get their head around because they think of a US like an open-source Claude Code. So what does that mean that their Slack bot was running on OpenCode? **Jay V** [22:13] Yeah, so you can think of OpenCode as a two-part product. There is the UI and the application part that you sort of see and that you interact with. But then there is the agent loop, the thing that's actually doing work while calling the LLM. And that is sort of behind the scenes. That's a little bit of like, we call it the server. That server can be embedded separately from the UI. And so in this case, they were embedding that and running their Slack bot off of that. **Jared Friedman** [22:41] YC's next batch is now taking applications. Got a startup in you? Apply at ycombinator.com/apply. It's never too early. And filling out the app will level up your idea. Okay, back to the video. **Diana Hu** [22:55] Can you tell us a bit about sort of these numbers and how the unit economics work? Because there was a shocking stat that was mentioned in Dylan Patel's podcast that Anthropic is profitable and by a huge margin. In Q2, they are on track to be doing $50 billion annualized revenue and around 70% margin. But they got there before the previous year was not profitable and way less. And they crossed this chasm, which sounds like where you're heading, but you don't have to subsidize it, which is special. **Jay V** [23:32] Yeah, I mean, we do subsidize a little bit,right? The subscription plan that we kind of have. But yeah. **Harj Taggar** [23:38] And you have a free tier too. **Jay V** [23:39] And we have a free tier, yeah. And I think that's the CAC part that you were sort of talking about early on where. **Diana Hu** [23:45] The new ads is rather than ads, the CAC is rather than paying for ads, is paying for tokens. **Jay V** [23:51] Yeah, it's because that's how people sort of experience that kind of magic moment. And that's the way we sort of get them into using the product, understanding what a coding agent is. All this stuff is definitely like, you know, a little too much for a lot of people. And I think the part here in terms of the unit economics that kind of works out is if you're actually paying per token there, in the case of Anthropic and in the case of how we sort of operate as well, we're able to get, at least in our scenario, we're able to get volume discounts because of the amount of tokens that we sort of serve. And so when you pay per token, that effectively turns into our margin. Whereas when you're subsidizing, yeah, obviously you're sort of eating the cost there. Or in the case of the free tier, yeah, you're eating the cost as well. But, you know, you were talking about this before, as you start to get more and more of these whales, those whales are paying, you know, per token and it's directly feeding into your margins. **Harj Taggar** [24:53] Yeah, the discounts you're able to get by sort of aggregating the tokens is interesting because, I mean, it's something we've talked a lot about over the last year in particular. It's where, I mean, everybody in the valleys at this point, where's the value going to accrue? Will it be the frontier models are going to make all of the money and everything at the app layer is left for dust? Or will it go the other way? How have you thought about that? Because, yeah, you're in an interesting spot because you're at your, you actually really do own the relationship with your end user and you're sort of effectively making it easy for them to pick and choose the models that they want to run with. So just what are your thoughts on where this all plays out and how it will hopefully play out in a good way for OpenCode? **Jay V** [25:33] Yeah, it feels like a little bit like a marketplace where a user is able to sort of make the choice of the model that they kind of want to use. And different models, different attributes, you know, different cost characteristics. And our claim here is, look, we want to showcase that diversity as well as sort of possible. And that also creates an environment where the labs are aware of each other and that competition ends up being good for the consumer in this case. Whereas the flip side is if you're locked into a specific vendor, then you don't benefit from, you know, the sort of competition that would otherwise sort of come with it. And you're likely, you know, helping their, maybe their margins in some ways. **Harj Taggar** [26:20] Yeah, I kind of feel like your growth is a pretty decent proxy for the fact that the choices only improved over time,right? **Jay V** [26:29] That's exactly it. Yeah, yeah. I think every, we could probably track back to sort of our bump in monthly active users down to some sort of bump in the open-source model market. And the way we sort of think about this is that it's not that we're picking a winner in terms of a model lab. We're just betting the field. We just think the rest of the field is going to do well. **Diana Hu** [26:53] The next logical conclusion of this is that models are becoming commoditized utilities. **Jay V** [27:02] Yeah, yeah. I think what's sort of interesting here is that the market is so large that it is hard to imagine people not, or model labs not picking off niches and chunks of their own in that they're specializing for specific areas or specific characteristics. Like in our case, when we sort of see some of the data that we were looking at before, you've got obviously the frontier labs, you know, we sort of know them well. But when we see the success of DeepSeek, it's very clear that they have picked the cost part of the quality, cost, performance sort of axes. And then they're basically saying, look, we want to be very, very good at that part. And, you know, it's hard not to imagine that happening across the board, across all these characteristics. **Harj Taggar** [27:49] Yeah, it feels like if you truly believe the sales pitch that the labs themselves make, which is, this is just an unprecedented market. Like the market for intelligence has not existed before. Everybody should be thinking in a positive sum, grow the pie mentality, in which case, like they should really want you to grow and succeed because you'll end up just being like a huge customer for all of them. **Jay V** [28:07] Yeah, and I think that that's sort of true now with these open-source models. We're the largest customer for most of them. **Harj Taggar** [28:14] You're the largest customer for most of the open-source models. **Jay V** [28:17] Yeah, yeah. Just in terms of the token volume, I think the amount that we do, yeah. **Diana Hu** [28:21] Wow. **Harj Taggar** [28:21] So that means that you guys must be sort of like locked in this like symbiotic relationship now where like you both need each other for this machine to work. **Jay V** [28:29] Yeah, the other half for us is the ecosystem,right? So we look at the open-source ecosystem as a whole and we're going, look, we need to make this entire thing sort of work. And again, with all the talk of sort of open-source models lately, that's essentially our pitch. **Harj Taggar** [28:43] Since you're the largest driver of most of the open-source model companies, you must have this incredible like insight into like the GPU market and where all this inference is actually happening because you're the ones driving the inference. Are you seeing anything interesting in the GPU market? Where's all this open code powered compute happening? **Jay V** [29:01] Yeah, so we rent GPUs. We work with providers that provide just straight inference. We work with the model labs themselves. But one thing that we started to notice that was sort of interesting at some point because of our global usage was that the peaks and troughs over the course of a day for GPU utilization wasn't that far off for us. Again, given the fact that like, you know, when sort of the east is sort of working, you know, maybe the west is sort of asleep, but when the west is working, the east is sort of, and so as a result, we have a reasonably stable 24-hour GPU cycle, which allows for pretty good utilization. And it helps sort of the unit economics for us in running these models a little bit more efficiently. And that ends up being a competitive advantage when we think about some of our counterparts that maybe just serve one part of the world. **Harj Taggar** [29:52] Do you think it's just such an interesting spot to be in? Like, do you think there are like specific product choices or design decisions you made when you were first launching the product that have led to just like quite like this sort of like the fact that you're growing and winning so much? ### Design **Jay V** [30:07] Yeah, it's funny, you know, the name OpenCode like literally comes from that. I think we've done a similar project in the past. It was called OpenNext. And the idea was when you've got a dominant or in this case, two dominant players in the market, the rest of the market coalesces around an open alternative. And picking that position ends up being really valuable because if you pick it, it's very hard for somebody else to displace you. And if it's open, you should try and become the default as quickly as possible. So when we had launched, you know, the name was a very deliberate choice. The fact that we wanted to support, even at the time of launch, we said we had claimed that we supported 70 plus models and providers. Just to make that happen, we had to create a separate open-source project called models.dev that built up this entire database that didn't exist at the time. And now obviously now you can contribute to it. And this is probably the best data set of all the models and providers out there in the world. But again, just to sort of make that happen and to, you know, occupy that position, it was a very deliberate choice. **Harj Taggar** [31:12] Can you think of any other examples of like intentional product or design choices you made that you think have really sort of helped you hit this inflection point? **Jay V** [31:20] Yeah, I think the name came a little bit later. This happened within a span of a few weeks. But yeah, the first thing I think that had happened was with our past product, we had just hit sort of being break-even. And around that time, this was February or so of 2025, Claude Code comes out. We look at Claude Code and we go, this is fundamentally different from using autocomplete, using AI in that form. And this is something that actually does make sense for us as a kind of a core developer. You know, we were sort of neo-Vim, Vim users at the time. And so Cursor didn't necessarily resonate with us as much. But watching sort of Claude Code and sort of using it a little bit, we kind of realized that it wasn't up to the standard of some of these other terminal UIs like the neo-Vims of the world. And we wanted to build something like that. And that was really the first sort of bit was that, you know, when you open this up, this should feel like a modern terminal experience to a lot of the core developer audience, the ones that we were going after very early on. It should feel like one of these other things. **Harj Taggar** [32:28] That's so interesting because I feel like, I feel like a lot of Claude Code users, because you have such low expectations for the term, like what you can get out of a terminal UI, people are like, wow, this is kind of crazy how much you can get done. And like these graphics and effects are really cute and cool and that's awesome. But like the fact that you were sort of like terminal UI connoisseurs, it sounds like you were looking at it thinking of, oh, like you could actually do so much more in the terminal. **Jay V** [32:50] Yeah, and we had built a couple of terminal UIs in the past, one as a part of our core product with SST. The other DAX I sort of built on the side with a couple of his friends. You could buy coffee online. It's a terminal UI. It's a complete storefront. It was sort of a fun pet project, but it showed off what we could actually do. **Harj Taggar** [33:10] Wait, wait, so your product was a terminal UI for buying coffee in case you wanted to buy coffee without leaving your terminal. **Jay V** [33:16] Yeah, because the idea was, you know, you're a hardcore developer. You're in the terminal all day. You can't be bothered to open up a browser,right? So you go ssh-terminal.shop and you order coffee over SSH. **Harj Taggar** [33:28] Yeah, it makes sense. Way easier. **Jared Friedman** [33:33] I feel like this is like an example of like of that PGSA where like if you're a developer and you build things that you want yourself, even if it seems really goofy to other people and a VC would like turn up their nose is like, well, that's a horrible business. What are you talking about? A terminal UI for buying coffee? It like has this tendency to like pull you in an interesting direction. **Harj Taggar** [33:54] Yeah, I just, it like generalizes, just like having eccentric tastes. It's like not always like leads to something interesting, but like it's kind of how you get to these like outlying things. **Jay V** [34:04] Because we were so embedded in the open-source community and because most of the people that sort of surrounded that community were people that looked up to products that were really good in the terminal, we knew that if we built one of those, it would resonate with them almostright away. **Harj Taggar** [34:18] It's pretty easy to think that you guys have sort of just came out of nowhere and have just exploded and are this like overnight success story. But the company's actually been around for a little bit longer than six months. ### Journey **Harj Taggar** [34:33] And so tell us a bit about that backstory because, you know, you went through YC over five years ago now, but even before that. **Jared Friedman** [34:41] Yeah, but the story starts way before even that. **Jay V** [34:44] Well, so it was second year university and I was like, I just done a co-op term at Waterloo and I was like, wow, I don't want to do this. And so I sort of come back, you know, being very naive, find the smartest people I can get a hold of. It ends up being Frank, my college roommate. And the two of us were like, yeah, let's just start a company. Again, I'm a 21-year-old and so I picked the name Anomaly as the name of the company because I'm like, I'm special. But yeah, it literally starts off with me reading PGSAs. **Harj Taggar** [35:14] How long ago is this? **Jay V** [35:15] 2006, 2007. And I think I started reading PG's essays, you know, because I was like, yeah, I want to start a company. I want to do a startup. And this was effectively, you know, the best thing you sort of find on it at the time. I think my first, the first YC application was probably around that time, but the first interview and what brought me the first time to Silicon Valley, I forget the exact batch, but I think this was the Mixpanel batch because I know the day that I interviewed was literally after Mixpanel, like Sueyala sort of interviewed. And I think this wasright, this wasright after Airbnb's batch. And so one of the Airbnb founders was hanging out, you know, in the sort of room when we were sort of waiting to be interviewed. And this was obviously with PG back then. And yeah, that was one. And then I think the first of many, first of many interviews and applications to YC. It took more than a decade to get in. I just put it that way. **Jared Friedman** [36:16] And you told me something crazy, which is I assumed that like during those years, these were like different startups. But apparently it was literally the same startup, like the same legal entity all of those years. **Jay V** [36:29] Yeah, yeah. **Diana Hu** [36:30] So this legal entity is almost 20? **Jay V** [36:33] Yeah, so we incorporated in 2010. **Jared Friedman** [36:36] So it's a 16-year-old legal entity. **Jay V** [36:39] Yeah, yeah. **Jared Friedman** [36:40] And you've had successful products before, but OpenCode is the most successful. And so it took 16 years since you started the company to have like a truly runaway success product. **Jay V** [36:49] Yeah, yeah. I think part of it is also to do with just sort of like your level of maturity, your level of, I guess, like ability, I guess. We were maybe too young for some of these other waves, like the mobile wave, the cloud wave, whatever, you know. And we did build things that did reasonably well. But this time around, it feels a little bit different in that it is a sum total of our experiences. And that's maybe made things a lot more easier to navigate, especially given the chaos in the space that we operate and the competition. **Diana Hu** [37:25] I looked at the numbers and you did nine applications from 2016 up to 2021 when you got accepted. **Jay V** [37:33] Oh, wow. **Diana Hu** [37:34] And you did four interviews. And for all of these, they were all different ideas, but it was the same legal entity. **Jay V** [37:43] It was the same legal entity, yeah. **Jared Friedman** [37:45] And the same founders. **Jay V** [37:46] And the same co-founders. **Jared Friedman** [37:48] It was Frankie and you and Frank doing it the whole time. **Jay V** [37:49] Yeah, doing the whole time. And then after we did YC, then our third co-founder sort of joined. And then it was the three of us for the last four years. It's just another chunk of time. It's just, you know, it's like flyaway. **Harj Taggar** [38:03] What was the idea you applied to YC with for 2021 that you got in with? **Jay V** [38:07] Yeah, so we were building a serverless platform at the time. It was like Heroku, but for AWS and serverless. And we wanted to do a better job in that space and maybe grow the market. And we built basically a serverless framework. That was our first big open-source project. And it was our first move into building open-source products, building in public, you know, doing the whole thing. And eventually, obviously, all of that sort of ties into OpenCode. **Harj Taggar** [38:37] Where does the building in public come from? Because you're clearly doing it now. You're like quite transparent with your metrics and growth, which is awesome. But yeah, where does that come from? **Jay V** [38:46] A little bit of that came from Dalton. I think he was sort of pushing us, just in general, just talking about like, yeah, you should probably do things in public because you're sort of an open-source company. Then I think it sort of dawned on us, this was maybe 2022 or so where it was like, it was actually Dax, one of our other co-founders, where the idea was just, look, you know, all your code is public. You work basically in public. If you don't talk about it publicly, you're probably doing yourself a disservice and your product a disservice. And, you know, ever since then, it's just been a part of our identity that, you know, for a lot of our community, the people that sort of follow us on Twitter, to them it feels, it sounds funny, but to them it feels like watching a reality TV show of this team and the company and the sort of journey that they're on. **Diana Hu** [39:34] I think the thing that's fascinating is that even though it sounds like such a windy road, and if people just heard the beginning of the podcast, the company sounds like a lightning in the bottle moment. It's like you just got caught by strike and got so lucky. But the reality is you've been grinding for a good 10 years and never gave up, which is so impressive. All those windy destinations that ended up being dead ends actually did teach you different things because you also ran a consumer company. **Jay V** [40:04] Yeah. **Diana Hu** [40:05] So you got really good at really all consumer acquisition, tracking numbers, and all of that, which really playsright now in the level of detail you have for OpenCode and of course open source and all of these. They were not all like wasted, quote unquote. **Jay V** [40:19] Yeah. **Diana Hu** [40:20] It was really more a journey that took 10 years to get to zero to 30 million in eight months. **Jay V** [40:27] Yeah, yeah, yeah. It's crazy when you put it that way. I think what's fun is that with OpenCode, we feel like we can go out and address the entire market. And that includes sort of consumers, individual users, to small teams, obviously the open-source community, you know, mid-market, all the way to the enterprise. And in our past iterations of all the different products we've worked on, we've probably done one of each. And so now it feels like, oh, we get to do all of them together in one product. And it's a lot more fun because you can think about the sort of customer journey through that entire thing. And it makes a lot more sense because you're not hyper-optimizing for specific parts,right? Like I'm not trying to build a very specific enterprise company or a very specific like consumer company. We're trying to just do the whole thing. **Diana Hu** [41:13] What got you to not give up? **Jay V** [41:15] Partly maybe being a little stubborn. It's funny. Maybe we should put a little warning that, you know, it's like don't try this at home. Because I think the prudent thing would have been, yeah, shut down your company, go join a high-growth startup, learn a bunch of things. But I think there was something in the back of my mind, Frank is probably the same, in that we felt like we were learning these different things along the way. And in our heads, we were sort of figuring out, okay, here's what it takes to build sort of not just a product, but, you know, do the marketing, to understand sort of the positioning of things, to do sort of the whole thing. And that journey felt like just progress and positive progress. And I think a part of that was obviously, you know, us being fortunate to have the ability to sort of do that, which was basically just living with parents for a bunch of time when we ran out of money. But yeah, yeah, maybe a little, a combination of being a little thick-headed and seeing positive progress. **Jared Friedman** [42:19] So I'll admit, I've been kind of a diehard Claude Code user since Gary Tan became addicted to Claude Code. But recently, I've been using OpenCode. And I dropped a bunch of PRs from OpenCode this week. And I've been really impressed with how well it works with open-source models on our existing code base, which is a very large, very complex code base. ### Try **Jared Friedman** [42:38] For folks who are watching who possibly have only used Claude Code or Codex or Cursor, how should they think about like trying OpenCode and potentially switching to it? **Jay V** [42:48] You know, when you hear about a new model that comes out, especially an open-source one, and you sort of want to try it out, you could kind of hack your way into using it with Claude Code or one of the closed-source alternatives. But OpenCode is really good for this. You just sort of go in, you look at the model picker and GLM 5.2 or whatever the new open-source model is, is probably up there. And you can pick it and start sort of using itright away. **Harj Taggar** [43:13] Okay, well, Jay, I think that's all we have time for today. I actually learned lots of really interesting stuff about your backstory that I didn't know in this episode. I mean, I think it's a pretty inspiring story, honestly, for anyone that wants to start a company. **Jay V** [43:26] Or at least entertaining. **Harj Taggar** [43:28] Entertaining and inspiring. **Jay V** [43:29] I know, it just hits on so many of the classic startup wisdom,right? Like you should pursue your interests, have eccentric tastes, be like live in the future a little bit. I would argue that trying to order coffee through the terminal is, I'm not sure if that's living in the future or the past, but it's like, it's not living in the current time. **Harj Taggar** [43:47] Maybe there's something in there. But yeah, I know, also just the fact that you kept like building your taste and just like to keep building things over a long period of time. And then when lightning strikes, like you're actually in a position to capture it. I think that's the thing that doesn't get mentioned actually, is like to catch lightning in the bottle, you actually like have to sort of position the bottle correctly and be ready for it and know what to do with it. Yeah, and you guys were well positioned to do that. So congrats on all your success. I know it's going to only get more explosive from here. **Jay V** [44:12] Thank you. Thank you for having me. --- This library is powered by PodHood (https://podhood.com), the podcast website platform.