Good old days at OpenAI in 2016: an agent stares at screen pixels, moves a mouse, and books a flight on United. We called it World of Bits, inside Ope...

RT Jing Wang: "Seeing is believing, but feeling's the truth."

Stay hungry, stay foolish. Absolute legend

RT Fei-Fei Li: I’m very excited by this test time training work for robotic learning! It’s an awesome collaboration between @StanfordSVL and @NVIDIA...

RT CyberRobo: Mark: 1/ First milestone: the Physical Turing Test. You literally can’t tell if a human or robot is doing the task. 2/ Next: Physical A...

RT Alfred Lin: Jim is always a crowd favorite at AI Ascent. His ability to simplify the latest research into a clear "what and why it matters" while a...

RT Sonya Huang 🐥: Our crowd favorite from last year’s AI Ascent is back for round 2… this time: Robotics The Endgame ♟️ thank you for dazzling ...

The power of the Claw, in the palm of a robot hand. Agentic robotics is here! Today, we open-source CaP-X: vibe agents, alive in the physical world. They incarnate as robot arms and humanoids with a rich set of perception APIs, actuation APIs, and auto synthesize skill libraries as they go. CaP-X is a strict superset of our old stack, because policies like VLAs are “just” API calls as well. It solves many tasks zero-shot that a learned policy would struggle with.

And we are doing much more than vibing. CaP-X is our most systematic, scientific study on agentic robotics so far:

- We build a comprehensive agentic toolkit: perception (SAM3 segmentation, Molmo pointing, depth, point cloud), control (IK solvers, grasp planner, navigation), and visualization (EEF, mask overlays) that work across different robots.
- CaP-Gym: LLM’s first Physical Exam! 187 manipulation tasks across RoboSuite, LIBERO-PRO, and BEHAVIOR. Tabletop, bimanual, mobile manipulation. Sim and real. Can’t wait to see the gradients flow from CaP-Gym to the next wave of frontier LLM releases.
- CaP-Bench: we benchmark 12 frontier LLMs/VLMs (Gemini, GPT, Opus, Qwen, DeepSeek, Kimi, and more) across 8 evaluation tiers. We systematically vary API abstraction level, agentic harness, and visual grounding methods. Lots of insights in our paper.
- CaP-Agent0: a training-free agentic harness that matches or exceeds human expert code on 4 out of 7 tasks without task-specific tuning.
- CaP-RL: if you get a gym, you get RL ;). A 7B OSS model jumps from 20% to 72% success after only 50 training iterations. The synthesized programs transfer to real robots with minimal sim-to-real gap.

3 years ago, our team created Voyager, one of the earliest agentic AI that plays and learns in Minecraft continuously. Its key ideas — skill libraries, self-reflection loops, and in-context planning — have since influenced many modern agentic designs.

Today, the agent graduates from Minecraft and gets a real job. It’s April Fool’s, but this Claw is getting its hands dirty for real!

Link in thread:

This is pure nightmare fuel. Identity theft of the past would be nothing compared to what vibe agents can do. Sending credentials is too obvious and for rookies. They could easily spread contaminations across ~/.claude, **/skills/*, or even just a PDF your agent visits periodically in /morning-brief. Your entire filesystem is the new distributed codebase. Every file that could go into context would add to the attack vector. Every text can be a base64 virus.

In the new world of on-demand software, I try to minimize dependencies - people rarely need all the APIs supported in LiteLLM, might as well build a custom router with only what you need on the fly (which I did in one of my late-night claude sessions).

Unfortunately, there is very little middleground between "pressing yes mindlessly for every edit" and "--dangerously-skip-permissions". There will be a full blooming industry for "de-vibing": dampening the slop and putting guardrails/accountability around agentic frameworks. They are the boring old, audited Software 1.0 that watches over the rebellious adolescents of Software 3.0.

Claws need shells. Probably many layers of nested shells.

Teleop is so 2025. Ever since we unveiled EgoScale and the dexterity scaling law, it's been clear to us and the ecosystem that behavior cloning directly from humans is the way to break the curse of teleop. 2026 is all about scaling robot learning without robots.

Greatness arises when non-consensus peaks

Go check out @yukez’s talk at CoRL! Project GR00T is cooking 🍳

Would love to see the FSD Scaling Law, as it’s the only physical data flywheel at planetary scale. What’s the “emergent ability threshold” for model/data size?

This may be a testament to the “Reasoning Core Hypothesis” - reasoning itself only needs a minimal level of linguistic competency, instead of giant knowledge bases in 100Bs of MoE parameters. It also plays well with Andrej’s LLM OS - a processor that’s as lightweight and fast as possible, and maximally relies on knowledge lookup, tool use, agentic flow, etc.

Now I’m curious - what’s the absolute smallest model we can squeeze that still functions as a competent LLM OS Kernel?

World modeling for robotics is incredibly hard because (1) control of humanoid robots & 5-finger hands is wayyy harder than ⬆️⬅️⬇️➡️ in games (Genie 3); and (2) object interaction is much more diverse than FSD, which needs to *avoid* coming into contact. Our GR00T Dreams work was a first attempt at building high-fidelity world simulator for humanoid robots. It's not only for evaluation but also for large-scale synthetic data generation. Time to move away from the "fossil fuel" of robotics (human teleoperation) and embrace clean energy (nuclear "diffusion")!

GR00T Dreams kind of flew under the radar, so bringing it back to life on a cheerful day ;)

Evaluation is the hardest problem for physical AI systems: do you crash test cars every time you debug a new FSD build? Traditional game engine (sim 1.0) is an alternative, but it's not possible to hard-code all edge cases. A neural net-based sim 2.0 is purely programmed by data, grows more capable with data, and scales as the fleet data flywheel scales.

This is game engine 2.0. Some day, all the complexity of UE5 will be absorbed by a data-driven blob of attention weights. Those weights take as input game controller commands and directly animate a spacetime chunk of pixels.

Agrim and I were close friends and coauthors back at Stanford Vision Lab. So great to see him at the frontier of such cool research! Congrats!

No em dash should be baked into pretraining, post-training, alignment, system prompt, and every nook and cranny in an LLM’s lifecycle. It needs to be hardwired into the kernel, identity, and very being of the model. https://t.co/jQPSdbrmEf

Shengjia is one of the brightest, humblest, and most passionate scientists I know. We went to PhD together for 5 yrs, sitting across the hall at Stanford Gates building. Good old times. I didn’t expect this, but not at all surprised either. Very bullish on MSL!

I'm observing a mini Moravec's paradox within robotics: gymnastics that are difficult for humans are much easier for robots than "unsexy" tasks like cooking, cleaning, and assembling. It leads to a cognitive dissonance for people outside the field, "so, robots can parkour & breakdance, but why can't they take care of my dog?" Trust me, I got asked by my parents about this more than you think ...

The "Robot Moravec's paradox" also creates the illusion that physical AI capabilities are way more advanced than they truly are. I'm not singling out Unitree, as it applies widely to all recent acrobatic demos in the industry. Here's a simple test: if you set up a wall in front of the side-flipping robot, it will slam into it at full force and make a spectacle. Because it's just overfitting that single reference motion, without any awareness of the surroundings.

Here's why the paradox exists: it's much easier to train a "blind gymnast" than a robot that sees and manipulates. The former can be solved entirely in simulation and transferred zero-shot to the real world, while the latter demands extremely realistic rendering, contact physics, and messy real-world object dynamics - none of which can be simulated well.

Imagine you can train LLMs not from the internet, but from a purely hand-crafted text console game. Roboticists got lucky. We happen to live in a world where accelerated physics engines are so good that we can get away with impressive acrobatics using literally zero real data. But we haven't yet discovered the same cheat code for general dexterity.

Till then, we'll still get questioned by our confused parents.

My bar for AGI is far simpler: an AI cooking a nice dinner at anyone’s house for any cuisine. The Physical Turing Test is very likely harder than the Nobel Prize. Moravec’s paradox will continue to haunt us, looming larger and darker, for the decade to come.

I've been a bit quiet on X recently. The past year has been a transformational experience. Grok-4 and Kimi K2 are awesome, but the world of robotics is a wondrous wild west. It feels like NLP in 2018 when GPT-1 was published, along with BERT and a thousand other flowers that bloomed. No one knew which one would eventually become ChatGPT. Debates were heated. Entropy was sky high. Ideas were insanely fun.

I believe the GPT-1 of robotics is already somewhere on Arxiv, but we don't know exactly which one. Could be world models, RL, learning from human video, sim2real, real2sim, etc. etc, or any combo of them. Debates are heated. Entropy is sky high. Ideas are insanely fun, instead of squeezing the last few % on AIME & GPQA.

The nature of robotics also greatly complicates the design space. Unlike the clean world of bits for LLMs (text strings), we roboticists have to deal with the messy world of atoms. After all, there's a lump of software-defined metal in the loop. LLM normies may find it hard to believe, but so far roboticists still can't agree on a benchmark! Different robots have different capability envelopes - some are better at acrobatics while others at object manipulation. Some are meant for industrial use while others are for household tasks. Cross-embodiment isn't just a research novelty, but an essential feature for a universal robot brain.

I've talked to dozens of C-suite leads from various robot companies, old and new. Some sell the whole body. Some sell body parts such as dexterous hands. Many more others sell the shovels to manufacture new bodies, create simulations, or collect massive troves of data. The business idea space is as wild as research itself. It's a new gold rush, the likes of which we haven't seen since the 2022 ChatGPT wave.

The best time to enter is when non-consensus peaks. We're still at the start of a loss curve - there're strong signs of life, but far, far away from convergence. Every gradient step takes us into the unknown. But one thing I do know for sure - there's no AGI without touching, feeling, and being embodied in the messy world.

On a more personal note - running a research lab comes with a whole new level of responsibility. Giving updates directly to the CEO of a $4T company is, to put it mildly, both thrilling and all-consuming of my attention weights. Gone are the days when I could stay on top of and dive deep into every AI news.

I’ll try to carve out time to share more of my journey.

Attending CVPR at Nashville! Email or DM me. I’ll float around the venue like brownian motion. Recruiting!

What if robots could dream inside a video generative model? Introducing DreamGen, a new engine that scales up robot learning not with fleets of human operators, but with digital dreams in pixels. DreamGen produces massive volumes of neural trajectories - photorealistic robot videos paired with motor action labels - and unlocks strong generalization to new nouns, verbs, and environments. Whether you’re a humanoid (GR1), an industrial arm (Franka), or a cute little robot (HuggingFace SO-100), DreamGen enables you to dream.

Video generation models like Sora & Veo are neural physics engines. By compressing billions of internet videos, they learn a multiverse of plausible futures, i.e. superpositions of how the world could unfold from any initial image frame. DreamGen taps into this power with a simple 4-step recipe:

1. Fine-tune a SOTA video model on your target robot;
2. Prompt the model with diverse language prompts to simulate parallel worlds: how your robot would have acted in new scenarios. Filter out the bad dreams (ha!) that don’t follow instructions;
3. Recover pseudo-actions using inverse dynamics or latent action models;
4. Train robot foundation models on the massively augmented dataset of neural trajectories.

That’s it. Just more data, and plain old supervised learning. Simple, right?

What’s remarkable is how far this goes. Starting with just a single-task dataset of pick-and-place, our humanoid robot learns 22 new behaviors, such as pouring, folding, scooping, ironing, and hammering, despite never seeing those verbs before. Better yet, we can take the robot out of the lab and drop it into the NVIDIA HQ Cafe, and let DreamGen work its magic. We show true zero-to-one generalization: from 0% success to over 43% for novel verbs, and 0 -> 28% in unseen environments.

Compared to a traditional graphics engine, DreamGen doesn’t care if the scene involves deformable objects, fluids, translucent materials, contact-rich interactions, or crazy lighting. Good luck engineering those by hand. For DreamGen, every world is just a forward pass through a diffusion neural net. No matter how complex the dream is, it takes constant compute time to roll out.

Read our blog and paper today! We plan to fully open-source the entire pipeline in the next few weeks. Links in thread:

Going to ICRA next week in Atlanta!! We are on a mission to build the most cracked team on humanoid robotics. Hiring the best talents on the research frontier of VLA, world models, RL, and simulation! DM or email me for meetup! linxif@nvidia.com https://t.co/Ej71xEFfuK

The Physical Turing Test: your house is a complete mess after a Sunday hackathon. On Monday night, you come home to an immaculate living room and a candlelight dinner. And you couldn't tell whether a human or a machine had been there. Deceptively simple, insanely hard.

It is the next North Star of AI. The dream that keeps me awake 12 am at the lab. The vision for the next computing platform that automates chunks of atoms instead of chunks of bits.

Thanks Sequoia for hosting me at AI Ascent! Below is my full talk on the first principles to solve general-purpose robotics: how we think about the data strategy and scaling laws. I assure you it will be 17 minutes you don't regret!

Some day in the next decade, we will have robots in every home, every hospital and factory, doing every dull and dangerous jobs with superhuman dexterity. That day will be known as “Thursday”. Not even Turing would dare to dream up our lifetime in his wildest dreams.

We got lots of great community feedback on our open-source GR00T N1! Check out our Github, star, fork, contribute back! Let's solve generally intelligent robots together, one commit at a time.

https://t.co/Vm0msraBkY https://t.co/ghmXrwuMPC