Kimi K2.5: The Best New Model is Open-Source (Again!)

Josh: Over the last few weeks, one of the hottest new topics that exists in the world of AI has been agents. Josh: Agents that can actually get into your computer and do things for you. Josh: We've seen this with Claude Cowork, which can actually access files and make Josh: changes on your computer.

Josh: And then to the fullest extent, we just recorded an episode on Claude Bot, Josh: which allows an actual computer to fully take over your life, Josh: send messages on your behalf, take care of emails, book reservations. Josh: The problem with that is that even an afternoon of use can cost hundreds of dollars.

Josh: So what we've done today is we've actually figured out a Josh: way to replace that totally for free where Josh: you get the same quality outputs but with none of the cost and Josh: to do that i'm going to start on anthropics website because the Josh: reality is that the screen that you're seeing right now isn't actually anthropics Josh: website in fact it was built using this new tool completely for free in about

Josh: 25 minutes which i thought was such an amazing demo it was built using a tool Josh: called kimmy k 2.5 which is the newest model coming out of china that is fully Josh: open source fully open weight and in order to build this all i had to do was Josh: feed it a video so you'll see on the screen here i Josh: generated a video on my desktop using a screen recorder that copied the Anthropic Josh: website, I said, hey, just take the screen recording of the website and create

Josh: an exact replica of the website for me. Josh: 25 minutes later, without any additional prompts, it listed all the things it did. Josh: It went through all the design and then actually published a full preview of Josh: the website that we can see here on this read record. Josh: So this model is incredible. I don't know if you've had a chance to play around Josh: with it or check it out, but this is a huge change in the world of agents because Josh: of how capable it is for such a low cost.

Ejaaz: This model is trending pretty heavily online right now. Ejaaz: I actually saw someone describe Moonshot Labs, the creator of this model, Ejaaz: as the anthropic of China.

Ejaaz: This was a quiet release, Josh. So the creators of this model kind of just updated Ejaaz: their chatbot interface with Kimi K 2.5 and didn't tell anyone. Ejaaz: And within a few hours of that launch, remember, no publicist or anything, Ejaaz: It was the number one trending model on Hugging Face, which is like where everyone Ejaaz: goes to access all these free open source AI models. And... Ejaaz: What you just demonstrated, I think, is one of the core reasons why this model is so special.

Ejaaz: So to give a few stats about this, it was trained on about 15 trillion tokens. Ejaaz: And typically, AI models are trained on text tokens specifically. Ejaaz: This wasn't the case with Kimike 2.5. It was trained on text and audio and visual Ejaaz: and a bunch of other mediums.

Ejaaz: And the reason why this is important is it allows you to do the example that Ejaaz: you just show, Josh, which is feed it a video or in this case, Ejaaz: a screen recording of a website that you wanted to build and build it in exactly that way. Ejaaz: And the reason why this is important is it shifts the use of AI models from Ejaaz: explaining what you need to do to it.

Ejaaz: So like, hey, could you do this, like describing what you want from it into Ejaaz: just showing it what you want to build. Ejaaz: And I think that that's like a really intuitive way for people to interact with Ejaaz: AI models versus like people that aren't just quite literate like me sometimes Ejaaz: when I'm trying to explain something. Right. Ejaaz: The second really cool thing is you started off this episode mentioning agents, Josh.

Ejaaz: And I think this is really important because KimiK 2.5 has this superpower where Ejaaz: they can spin up up to 100 sub-agents. Ejaaz: Think of a sub-agent as just another instance or replica of KimiK 2.5, Ejaaz: but it's specifically focused on doing a certain task. Ejaaz: So for example, if your goal is to figure out whether investing in Anthropic is a good idea.

Ejaaz: It'll spin up one agent that does the research, another agent that does the Ejaaz: fact checking, another that tests different kind of architectures.

Ejaaz: And the cool part about this is they can work in parallel, which means that Ejaaz: you can cut down the execution time for a task by four and a half times. Ejaaz: So imagine you had a task that took four and a half hours, you can now do it in one hour.

Ejaaz: And I think this kind of like multi-agent trend that you identified or that Ejaaz: you spoke about is super important because that's what we're seeing with the Ejaaz: likes of Anthropic with Cloud Code and Cowork and OpenAI with Codex. Ejaaz: But the fact that this thing is free is completely insane. Josh, Ejaaz: do you know how much it cost, rumored, for them to train this?

Josh: I have no idea, but I would imagine a tremendous amount of money for them to Josh: just train it and then release it fully open source, open weight. Ejaaz: So the rumor, and again, this is not fact. I wish I could fact check this. Josh: And also, to be fair, before you say this, the Chinese models are notorious Josh: for lying about how much it costs. Yes, they are. So take this number with a grain of salt. Ejaaz: You're right. So the number that's being floated is $4.6 million,

Ejaaz: which is nothing. That seems so low. Ejaaz: It seems so low, which is nothing compared to the billions of dollars that OpenAI Ejaaz: has spent to kind of train their models. Ejaaz: And to give you guys an idea, like why we're comparing Kimi-K 2.5 to these like Ejaaz: frontier AI models built by OpenAI and Anthropic is because in some cases, Ejaaz: it's almost as good as this.

Ejaaz: Like if you look at its performance on humanity's last exam, Ejaaz: which is notoriously the hardest benchmark for an AI model to be tested on, Ejaaz: it scored a 50.2%, which beats Claude's latest model, Opus 4.5, and GPT 5.2. Ejaaz: It doesn't quite beat Anthropic at coding, Josh.

Ejaaz: I know you built that cool website in a few minutes, and which makes me think Ejaaz: that maybe it's really good at front end development but just a really impressive Ejaaz: model and i'm guessing it's like super cheap to operate as well compared to

Ejaaz: like some of these expensive. Josh: Models yeah we're going to get into the cost because if you do want to use Josh: it at length and you don't have a couple h100 gpu sitting in your your home Josh: you're going to have to pay a little bit uh thankfully it's significantly less Josh: and we'll get into the prices but one thing you mentioned is that it's actually Josh: not the best coding model in the world and i think that's okay that's not the

Josh: real breakthrough one of the most amazing breakthroughs is actually before we Josh: were recording the show, Ijaz, you showed a demo of, Josh: you gave CloudCode the website of Figma and said, hey, can you go emulate this Josh: website? And it actually did a pretty good job. Josh: The difference between CloudCode and something like this new model is that

Josh: I was able to feed it just a video. And what it did is it analyzed each frame, Josh: each pixel within each frame, understood the context of each pixel, Josh: and then figured out how to intuitively regenerate that in a webpage using code Josh: and like whatever type of design tools that it used. Josh: And that is the novel thing because most models do image to code, Josh: but Kimi K 2.5 does video to understanding to code. And I think that's one of Josh: the more novel breakthroughs.

Josh: One of the three, actually. The second with it being natively multimodal. Josh: So you mentioned 15 trillion tokens that it was trained on, but it's mixed between Josh: visuals and text for the first time. Josh: So this really has a good understanding of videos, of photos. Josh: It's starting to even get the Google physics. Josh: And then the third part that she mentioned, which is the Asian Swarm. Josh: I want to spend a little bit of time on this because the Asian Swarms are super cool.

Josh: We actually have an example of one of the Asian Swarms and how it works. Josh: The way it works is it's able to separate...

Josh: Itself into basically a hundred small mini tasks and the example that Josh: we're seeing on screen now is a film script and Josh: it's a short story that the model generated it created Josh: a shot list it created renderings of images of Josh: what this the frames of this could look and it generated basically Josh: an entire movie in a fraction of the time that it Josh: would take to do as a single model i think the Josh: actual number is four and a half times faster like

Josh: you mentioned earlier versus traditional models so it can call up to 1500 tools Josh: it's like this swarm of agents working on a single problem so it's faster it's Josh: more efficient it can it's just like you can make a movie script in five minutes Josh: and it'll generate the entire thing for you with a shotless and all i mean some Josh: of these examples are pretty unbelievable did Ejaaz: You hear about the the underlying mechanism that they use to to build this it's

Ejaaz: actually super cool because well one thing chinese ai model labs are repeatedly known for doing is, Ejaaz: you know, they don't get access to all these fun, expensive GPUs that the Western labs get to. Ejaaz: So they have to get really creative in their research and training techniques. Ejaaz: And they did that with the agents. Ejaaz: So to get that four and a half times efficiency that you mentioned, Ejaaz: they use this technique called parallel agent reinforcement learning.

Ejaaz: Typically, when you spin up 100 agents. Ejaaz: You're going to have a hard time. And the reason why you're going to have a Ejaaz: hard time is something called agent collapse. Ejaaz: So typically, a model will be used to doing things in sequence. Ejaaz: So if you ask it to do a really complex task, it's going to start with task Ejaaz: one and only proceed to task two once it's done with task one. Ejaaz: And if you spin up a bunch of agents, the model might sometimes still do things

Ejaaz: sequentially. And you don't want that. You want it to do in parallel. Ejaaz: So this new training technique that they spun up and pioneered, Ejaaz: there's a paper all about this, is super unique and never been done before, Ejaaz: which allows them to not get model collapse at 100 agents.

Ejaaz: The crazier part is each of these agents get access to over 1,500 tools. Ejaaz: And that's what makes an agent useful. You go from an LLM telling you what is Ejaaz: useful to an agent that can actually do something on your behalf. Ejaaz: That's pretty impressive. And then the final thing is they have this thing called, Ejaaz: well, I actually don't know what it's called technically, but the way I understood Ejaaz: it is it's kind of like a brain. It's called an orchestrator.

Ejaaz: And so it ingests the tasks that you've asked it to do, and it breaks it down Ejaaz: into multiple different tasks. Ejaaz: The fact that it could do it for this cheap, Josh, sorry, I know I keep mentioning Ejaaz: the cost, but you need to tell the people the cost because it's just insane. Ejaaz: This is something that I would use regularly, a company would use regularly Ejaaz: because it saves them so much money.

Josh: And we actually have this really cool visual of the orchestrator on screen here, Josh: which gives you a visual representation of what that looks like. Josh: So the orchestrator breaks this down into sub-agents. It assigns them tasks. Josh: And then the tasks kind of go back and forth through a fact checker. Josh: There's a file downloader. There's a web developer. There's this entire toolkit. Josh: So one of these is emulating an AI researcher. The other is a physics researcher.

Josh: The other is a life science researcher. Josh: And what you're getting is a series of experts across every domain working on Josh: problems in parallel with access to 1,500 of these tools, like the fact checker, Josh: the file downloader, the web developer, Josh: the text scraper, so it can view images and understand and interpret what they Josh: mean. And it's such a powerful...

Josh: Stack that you have. And without the collapse, with this software novelty that Josh: they've introduced, it allows them to do this unbelievable thing. Josh: So to your point, when, I mean, historically, China has been hardware constrained Josh: and they've really accelerated on the software.

Josh: And this is very much an extension of that acceleration. Now we have some more Josh: examples that are very fun to show that I would love to show because as I was Josh: going through to prepare for the episode this morning, I was like, Josh: wow, this is pretty cool. Josh: And EJ, you even dropped in one of your own that I thought was pretty neat. Josh: So if you don't mind explaining what this one is here.

Ejaaz: Well, it may not be the coolest example, but this is something that I would personally do. Ejaaz: And I know a bunch of my friends would do in their spare time, Ejaaz: which is just again, on websites, the fidelity of these things is pretty insane. Ejaaz: And I have to emphasize, we're going from a screen recording to like a fully Ejaaz: functional front end development. Ejaaz: And I don't think people quite understand how necessarily hard it is to do front end development.

Ejaaz: I think a lot of software engineers that do this will kind of scoff at that comment. Josh: But it is true. Ejaaz: It is like super hard to do because there's the design element, Ejaaz: which is incredibly subjective, which is Kimi K2.5's exact point. Ejaaz: Instead of trying to describe it to an LLM, you can just kind of take a screen Ejaaz: recording and spin this up in a matter of minutes, right? It took you, I think, 7.5 minutes.

Ejaaz: I just want to like emphasize a point that you made earlier before this, Ejaaz: Josh, which is the agent side of this model is super important because if you Ejaaz: look at a model from Anthropic, Ejaaz: their flagship product is Claude Code and recently Claude Cowork. Ejaaz: If I were to tie both of those products in one unique trait, Ejaaz: it's the fact that you can spin up multiple agents.

Ejaaz: In fact, the founder of Claude Code, and in fact, a lot of the Anthropic team Ejaaz: do between 80 to 100% of production level code. Ejaaz: So that means new products that they ship completely built by Claude Code. Ejaaz: Now, they're not doing this using one model.

Ejaaz: They're doing this spinning up several instances of. Ejaaz: Kind of like put this into perspective this is Ejaaz: the future of software development and software development pretty Ejaaz: much underpins any major major breakthrough for Ejaaz: any industry going forwards software and tech underpins everything Ejaaz: so if you have an ai model that costs a fraction of the amount that the frontier Ejaaz: flagship model from anthropic does and is 100 free and open source yeah you

Ejaaz: might need whatever 50k to 100k's worth of gpus to run it on your own instance Ejaaz: but you can get access to Kimi K2.5's API right now. Ejaaz: That is a huge advantage. And the Chinese AI labs just somehow stay on top. Ejaaz: I don't know how they do this. Josh: Yeah, well, if we're talking about token price, maybe we could get into the economics first.

Josh: We'll skip ahead a little bit because I think the economics of this is super Josh: interesting, where if you don't have access to the GPUs in your house, Josh: which I'm guessing nobody listening to this episode really does, Josh: if you don't work at a major AI lab, well, there are ways in which you can run Josh: this for free or close to free.

Josh: Now, the example that i showed earlier where i created a website clone Josh: that was free because kimmy gives you three agentic Josh: tasks per week basically that you can use for free but after you've exhausted Josh: that there are some economic pricing sheets that we can use to kind of compare Josh: this to other models in terms of cost so for opus 4.5 which is the most popular Josh: model that we've been using everyone's been using it's fairly expensive the price input

Josh: Per token per million tokens is five dollars Josh: while the output is 25 dollars so for Josh: every million tokens you generate with opus 4.5 which is Josh: the flagship coding model it costs 25 bucks for kimmy k 2.5 the input is 60 Josh: cents per million tokens and the output is three dollars that is almost a full Josh: order of magnitude 90 decrease in price relative to opus 4.5 at a very comparable rate.

Josh: And that's just if you're comparing it on code. If you're comparing it on general Josh: agentic tasks, it's actually slightly more capable than Opus 4.5 for one-tenth of the cost. Josh: So if you're using something like CloudBot, which we recorded an amazing episode Josh: on earlier this week, which you should go check it out, you can just swap in Josh: this new model and run it through whatever cloud service you want. Josh: And the price of your agent will be one-tenth of the cost.

Josh: And this happened in the matter of a couple of days. So the costs are rapidly decreasing.

Josh: And I think that advantage of it one being open source but two Josh: being cost effective is huge for everyone if you remember Josh: i think it was two or three weeks ago anthropic cut off Josh: xai from using cloud code they actually removed Josh: access to it and because it's closed Josh: source there's nothing they could do about it but if they're using a model like kimmy Josh: k2 to run k2.5 to run Josh: their agents to build their code there's no one who can actually sever that

Josh: tie and it's the same for developers where if you're building on a platform Josh: and you don't want it to change well now you have the open weights it's going Josh: to be locked in forever it's going to be a fraction of the cost this is a really Josh: viable substitute for those who are loving the agentic life os workflows Ejaaz: Do you have any idea how they're able to produce this for such cheap costs?

Ejaaz: Because I'm trying to rack my brain around this, right? So, okay, Ejaaz: sure, you've made a few research breakthroughs. Ejaaz: The Chinese labs in particular are known for discovering or commercializing Ejaaz: mixture of experts, which kind of cut down prompting and inference costs and Ejaaz: training in general to a fraction of the price. Ejaaz: But still, they don't have access to some of the top hardware, right?

Ejaaz: And kind of like Moore's Law would state that eventually a bunch of these GPUs Ejaaz: that are A-grade are going to Ejaaz: cost so much less and run like 100x more inference performance per token. Ejaaz: So it's going to cost a lot less in general over time. Ejaaz: They don't have access to these resources that the West does, Ejaaz: right? So you mentioned like Anthropics cost, right? Ejaaz: You said what? It was like $5 in and how much out? Josh: $25 per million tokens compared to three.

Ejaaz: Okay, that used to be $15 in and $75 out when they launched the product. Ejaaz: So we've come down by a significant factor since then. But again, Ejaaz: I would imagine they did this because of cheaper, more effective chips. Ejaaz: How has Kimi K2.5 done that? How has Moonshot done that? Josh: I think I have two answers to this question. The first is through software innovation.

Josh: I assume they have cracked some sort of a code that allows them to generate less tokens per output. Josh: The second one is the margins. Ejaz, how much money did Anthropic make last year? Josh: It was what, $10 billion of revenue? Ejaaz: $10 billion, correct. Josh: China is unfortunately not the leader in AI and therefore they need every incentive Josh: in the world to dethrone the leader of AI.

Josh: One of the ways you could do that is by winning on the margins and cutting down Josh: those margins for your competitors. Josh: Clearly they have this strategy because they're publishing this open source, Josh: open weight. And you're seeing that happen with the pricing as well. Josh: I assume a large part of that revenue Josh: from Anthropic is just margin on the inference that they're charging.

Josh: It doesn't cost them anywhere near $25 per million tokens, but they're able Josh: to charge for it because they're the leading frontier model that all of these Josh: labs and businesses are willing to pay in order to use their services. Josh: In the case of Kimi K2.5, they don't care. They don't need to make profit. Josh: They just want them in market share. Josh: And to do that, they're able to undercut pretty aggressively here. Like I'm sure...

Josh: Anthropic could match this and perhaps not actually lose money. Josh: But that profit thing is real. Ejaaz: It also helps that they have an absolute gigabrain as their founder and CEO. Ejaaz: I don't know if you've looked into this guy, but this dude is only 31 years old. Ejaaz: He was born in China. He went to Tsinghua University, which is actually the Ejaaz: most popular university for AI and ML researchers in the world to graduate from.

Ejaaz: 50% of the world's top AI researchers, by the way, reside in China, Ejaaz: and a large chunk of them graduated from Tsinghua. Ejaaz: But Josh, he also did his PhD at Carnegie Mellon and he did it in under four Ejaaz: years in assumedly robotics and machine learning, which is very impressive. Ejaaz: And he also did a very long stint building out Google Brain and meta AI research. Ejaaz: So he was probably one of those meta researchers getting paid tens of millions

Ejaaz: of dollars a year. So this guy's track record is insane.

Ejaaz: So it doesn't, I guess, with that CV, doesn't kind of surprise me that he's Ejaaz: made these breakthroughs somehow. even on the hardware that he's constrained. Josh: Yeah, it's incredibly impressive. I'd love to hear more from them. Josh: In fact, we actually, the first time we heard from the founder was earlier today Josh: with the announcement post. I had never really seen what he looked like. Josh: I hadn't really heard him communicate.

Josh: It feels like it's a very sheltered, kind of quiet, secretive workplace that they have there. Josh: But I'm hopeful that we'll start to see more because my God, Josh: the talent there must be unbelievably impressive. Josh: Just in China in general, when we talk a lot about the trading competitions Josh: that we have, China's always seemingly winning.

Josh: They're just, they're doing really well and clearly they have incredible talent Josh: density now you're showing on screen something that i'm very excited to talk Josh: about which is jumping back to the examples of what you can actually do with Josh: this new model and one of them is this really fun blueprint to 3d model space Josh: now ijez you've watched friends right this might look familiar to you oh Ejaaz: Yeah uh not not the uh the one on the left yeah not the exact uh high.

Josh: Fidelity you don't know the blueprint of the Josh: Yeah, it's pretty cool. So it took a two-dimensional blueprint of a room and Josh: it generated a three-dimensional version of Monica's apartment or Monica and Rachel's apartment. Josh: I haven't watched Friends, but I know it's very popular and I've seen clips Josh: from this room. So I'm familiar which one it is.

Josh: And it's a testament to the types of new creative things that you can do now Josh: that it has the image to critical thinking to output. Josh: Uh type of thinking process through generating these outputs and i just thought Josh: that was really interesting there's a lot of really fun use cases that you can use and Ejaaz: Dude this is a this is a ten thousand dollar a month apartment at minimum josh Ejaaz: it's it's making me feel poor looking at the schematic oh my god yeah right right.

Josh: Growth street new york city that might even be more than 10 grand that's prime real estate come Ejaaz: On dude yeah that's insane how how were they able to pay rent they were making Ejaaz: comedic jokes the entire time for for seven years.

Josh: But also this becomes a very useful tool for real Josh: estate agents right because they want to kind of recreate spaces Josh: allow you to feel and live in the space more and Josh: granted this is a low fidelity version but i'm sure this is step one in Josh: creating some higher fidelity mock-ups of spaces that you would possibly want Josh: to rent if you're building a house if you're building anything this is great

Josh: for construction for modeling these services used to cost a ton of money for Josh: virtual renderings now they're effectively free or very close to it maybe just Josh: a couple cents per output and that decrease is pretty substantial open

Ejaaz: Source is having quite the. Josh: Week they're having a moment i've Ejaaz: Commented a lot about this before but um i've said Ejaaz: that i i never think open source will actually ever catch Ejaaz: up to frontier level uh capabilities and in this case in some ways it does in Ejaaz: some ways it doesn't um josh you know uh in my period or era of life right now Ejaaz: i am a coding agent maxi i'm incredibly bullish on anthropic so uh you know

Ejaaz: i i scrutinize any other competitor pretty heavily when it comes down to this. Ejaaz: I don't think it is as good as CloudCode. You mentioned this earlier, Ejaaz: but it's scarily good in some aspects, right? With the front end development. Ejaaz: So I'm curious to see how people use this. And I think what I love most about Ejaaz: this is a lot of my friends that kind of want to do more creative pursuits, Ejaaz: like build websites and do more front end stuff.

Ejaaz: They don't want to pay 200 bucks a month, CloudCode max, right? Ejaaz: But they can get this for free and they can access it today. Ejaaz: You literally built your website today, like in a few minutes before the The Ejaaz: show starts and then recorded it. Josh: In 25 minutes with one prompt.

Ejaaz: That's that's insane that's insane so if you can do it if i can do it anyone Ejaaz: else listening to this can do it definitely go give it a go like i want to see Ejaaz: some examples that people kind of like do with this.

Josh: The kimmy website itself actually has a bunch of Josh: ideas and use cases that you can use to kind of Josh: emulate or get inspired by and this is one of the major things Josh: with the model launch like the reason we're talking about this today is Josh: because they provide this really awesome demo online of them Josh: screen recording a website and then emulating that and Josh: creating it in five minutes and that's what we did and that's why it's so

Josh: exciting so the models that are not only able to make it accessible Josh: through lower cost pricing but to kind of give you these curated Josh: experiences where you can satisfy some sort of goal that you want in a way that's Josh: easy all i asked was hey just create a clone of this website do it identically Josh: and don't make any mistakes and it did it in one shot i think that is a critical Josh: threshold required to onboard a lot more people to be excited to use this stuff

Josh: to go and set up quad bot it's pretty technically challenging Josh: It takes a little while. It's not for the faint of heart. But something like Josh: this, where they give you these use cases, they make it available for basically Josh: free Mium, where you can pay extra if you want to use it more. Josh: It's really exciting to see. And yeah, I mean, China's totally having a moment.

Josh: And open source is totally having a moment. Like both of these things are converging Josh: at once to create all of the news this week, while the major AI labs who are Josh: closed source are just kind of working in silence, perhaps trying to figure Josh: out how to best react to something like this that becomes open source and available.

Josh: Now, you have to imagine, EJS, it's been a little while since we got a new big Josh: dog on the block, a new Frontier model for one of these labs. Josh: So the silence is deafening, but generally, the longer the silence goes, Josh: the bigger the boom that follows. Josh: And I suspect we are only a few weeks away from some new models that will make Josh: this Kimmy K2.5 look like child's play, which is crazy to see.

Josh: Because right now it feels unbelievable and magical, but I'm sure it is soon Josh: to be dethroned when the new models come out. So I guess we'll be here to follow Josh: along with all of that news. Josh: Ejos, any final thoughts before we part today? Ejaaz: I mean, once again, the clear winner for all of this is the users. Ejaaz: Big time. We get access to all these frontier models for either a cheap or free Ejaaz: option. It's super cool.

Ejaaz: Or if you want to pay the extra amount and get like a curated experience, Ejaaz: you can also do that as well. Ejaaz: If you want to use a Chinese AI model, go for it. If you want to use a Western lab, pick your poison. Ejaaz: What I'll finish up with is the pace of development for these things, Josh, is Ejaaz: So underrated. Like, I feel like we are so spoiled. When we first started this Ejaaz: show around like eight months ago.

Josh: We were like, oh man, Ejaaz: Like it can produce a pretty good market summary of this investment, Ejaaz: but like it's nothing like crazy. Ejaaz: Fast forward to today and I'm reading a tweet on my timeline from the founder Ejaaz: of Claude saying like, yeah, 100% of the code that we make, aka every new product Ejaaz: that we build going forward is managed by Claude, like is managed by Anthropic.

Ejaaz: And I can assume that with a product like KimiK 2.5, they're probably doing the same thing. Ejaaz: So are we entering the era where AI just builds itself? Probably. Ejaaz: Super scary. I read an essay last night, word to the wise, don't read scary Ejaaz: essays at night, where Dario Amode, founder of Anthropic, wrote about his bearish Ejaaz: thesis and why we need to be super careful going forwards because we're entering AGI, dare I say.

Ejaaz: I don't know. I'm super excited. These model developments are super cool.

Ejaaz: And I'm excited for Josh Codex is probably going to come up with an upgrade Ejaaz: OpenAI's new coding model is coming out in the next couple of weeks I'm excited Ejaaz: they're having a town hall today fingers crossed that they probably want to announce it but maybe Ejaaz: but when they do this will be the first platform to hear it on, Ejaaz: now I know a bunch of you have listened to this and are thinking hmm, Ejaaz: I'm going to download Kimmy K2.5 or just use it and test it out.

Ejaaz: I have a task for you to try out. Ejaaz: In fact, it involves not one, but two sub-agents. Ejaaz: Number one, ask it what the top AI show is on YouTube or any favorite platform Ejaaz: that you listen to or hear on. Ejaaz: And then ask it to subscribe if you aren't. Turn on notifications and give it a five-star rating. Ejaaz: I have asked this of you for the Claude Clawbot episode. Ejaaz: I'm going to ask it for you for any of the Kimmy K 2.5 fans out,

Ejaaz: please support us. It helps us massively. Josh: Yeah, if you ever need a use case, you can just have it. Go figure out how to Josh: subscribe autonomously to the YouTube channel. Ejaaz: That'd be pretty cool. Josh: And share it with your 10 closest friends through iMessage once you get hooked Josh: up with ClawBot. That would be great.

Josh: But yes, all of these cool, exciting new things that you were talking about, Josh: including Dario's Anthropic Letter Josh: and the OpenAI State of the Union that they're kind of hosting today. Josh: We're going to cover that on our episode later this week in the AI Roundup. So stay tuned for that. Josh: And yeah, we'll see you guys in that episode. Thanks for watching.