Kawasaki CORLEO 1.0 AI Robot Beast Demos Next Gen Collab Tech ($29999 HUMANOID)

And this thing is unlike a motorcycle in that, because it has four robotic legs, it can tackle uneven ground like forests or muddy riverbeds or rocky mountainsides, places where regular ATVs probably wouldn’t cut it, and this opens up brand new possibilities for exploration, according to Kawasaki. And you ride it like a horse. Apparently you just climb on and you huddle, and Kawasaki says that it’s beginner-friendly, so you don’t need to be a pro to get the hang of it, because it uses an AI system for its control, and instead of using handlebars or a throttle, it responds to how you move your body.

So you lean forward, and you shift your weight, and it adjusts and steps forward instantaneously. This makes it like the robot is reading your mind, as though where you move with your posture to the left or to the right, front or back, it will move you in that direction, like a hoverboard, but with four legs. And it runs on a hydrogen fuel cell that’s paired with a 150cc engine, which means that there’s no gas guzzling, and its built-in AI keeps it balanced and navigates from point to point, so it’s smart enough to handle the terrain without you micromanaging it step by step.

And there’s a heads-up display or a HUD that sits right in front of the machine, showing you stuff like how much hydrogen is left, where you’re headed, and how much weight is being distributed, etc. It’s almost like a video game interface that they’ve designed. And Kawasaki is playing the long game here, because although this is still a prototype, they’re saying that they’re trying to release this thing within the next 25 years. And the goal is to swap out dirt bikes for these robotic beasts, according to Kawasaki. And because clean energy is the new trend, Kawasaki released this Ninja H2 HYSC last year, and apparently this Corleo is building on top of that model.

And it takes a video game influence. According to the company, Corleo was inspired by Horizon Zero Dawn, which is a PlayStation 5 game with a giant robotic animal in it. But since it’s still in this prototype phase, Kawasaki hasn’t shared any specifics. So we don’t know yet exactly how much this robot’s going to cost, but we do probably know that it’ll at least have 12 degrees of freedom, with 3 degrees of freedom in each leg. But there’s still no word on top speed, range, battery life, or price, or release date. And make sure to comment down below how much you think this thing should cost, and when do you think it’ll actually come out, or who do you think will put this out first if it’s not Kawasaki.

But there’s another humanoid that’s about to enter your home with a brand new AI update, and it’s the Neo Gamma completing tasks 100% autonomously now, all running on a single set of neural network weights. And in the video, we can see Neo picking up leaves and bagging them all on its own, which is a small task, but it serves as a huge hint as to what’s coming, which are completely generalizable robots that can handle everyday chores autonomously. And One X is using a visual manipulation model that’s super flexible, which they’ve also paired with their in house language model so that Neo can talk to you naturally, and even throw in some body language while doing it.

So for instance, you can say, Hey, Neo, go clean up those leaves, and it’ll just do it. But that’s just the beginning because in theory, the same neural network we talked about can also load your dishwasher, fluff your pillows on the couch, tidy up, deep clean and manage your home, all while learning from the new data that it receives. And just remember, it’s one neural network that’s running Neo’s legs, arms, head or so all at once. And because One X is tackling the home market first, they’re calling this the final boss of robotics.

And this is because your house is totally unstructured. Every home is different. And even something simple like doing laundry could come with a million different subtasks like unscrewing the detergent cap, unbuttoning the shirts, folding stuff up just right. It’s all chaos. But the twist is that the chaos here is gold, because the home is where you strike the ultimate treasure, the super diverse data sets that One X is looking for. And this is the secret sauce for building a robot with general intelligence because you need tons of data from all kinds of tasks in weird environments to make robots that can handle anything.

And this is unlike many robotics companies that are trying to crack autonomy one task at a time. In this case, they master one thing and they make some cash and then they move on because autonomy is super tough. But One X is going straight forward instead. And they’re making great progress because Neo is quickly becoming a jack of all trades. And as for its features, so far, it has this emotive earring designed to communicate better with people. And the robot features a natural human like gate. And it’s designed with safety and mind featuring soft covers to make it more gentle while working inside of your home.

And it features this visual manipulation model to let it learn how to grasp all kinds of objects, even in places inside of the home that it’s never encountered before. And next, we have a brand new female humanoid robot from Cherry Imoga robotics. And they’ve teamed up with deep seek to boost their robots voice interaction capabilities. And they’ve integrated the robot with deep seeks v2 v3 and r1 AI models for more natural conversational abilities. And this is building on 15 years of AI work, including their lion AI large language model and Cherry says that they see deep seeks open source approach using Nvidia’s H 800 chips as the perfect fit for their tech.

And real world trials of the Imoga hardware are starting soon. And they’re going to test their performance outside of the lab. And Cherry’s AI journey actually began in 2010. As they hit a milestone with Cherry Lion in 2018 and launched their Imoga robots globally in 2023. And in 2024, they actually invested 100 million RMB for mass production of the robot. And now they plan to start selling it overseas midway through this year. And while the price of this robot is not yet known, their next goal, they say is an intelligent mobility ecosystem, blending their cars, robots and internet products together.

But make sure to comment below and tell us if you think this is just a marketing stunt or if it’s an actual humanoid product that’s heading to home soon. And finally meta just unveiled its llama for herd, which is a suite of natively multimodal AI models that are designed to deliver personalized experiences. And leading the pack are llama for Scout and llama for maverick. Both of these are 17 billion parameter models and they’re available for download on llama.com and hugging face right now. Plus users can also try these on WhatsApp, messenger, Instagram direct, and meta.ai.

And when it comes to size, llama for scout has 16 experts built in and it fits on a single Nvidia h 100 GPU. And it features a 10 million token context window, which outperforms both Gemma three and Mistral 3.1. On top of this llama for maverick features 128 experts, and it beats GPT 40 and Gemini 2.0 flash on benchmarks, offering a leading performance to cost ratio and an ELO score of 1417 on LM arena. And both llama for models leverage a mixture of experts architecture for efficiency. And they stem from llama for behemoth, which is a 288 billion parameter Titan that’s still in training and it surpasses GPT 4.5 and Claude sonnet 3.7 on stem benchmarks and metas open weight approach for Scout and maverick marks a first for multimodal mixture of experts models in the llama ecosystem.

And these llama for models are built for developers and enterprises, meaning that these models are designed to excel in reasoning, coding and multimodal tasks, including robotics with Scout offering industry leading context length and maverick rivaling deep seek v3. And meta is emphasizing its openness with its open source nature, planning on sharing more at the upcoming llama con. Anyways, like and subscribe. Thanks for watching. And if you want to know more about the latest in robots, check out this video here. [tr:trw].