Humanoid robots from Figure, Unitree, Tesla Optimus and others, plus robot arms, embodied AI and self-driving cars.
Agility Robotics just listed on Nasdaq via a SPAC merger at a $2.5B valuation, making it the first U.S. humanoid pure play. But with just $1.8M in 2025 revenue and over half coming from a single investor, Digit must now survive the scrutiny of quarterly earnings to prove warehouse humanoids can turn a profit.
Researchers from Tsinghua and UC Berkeley just open-sourced uni-robot-agent, a unified harness that lets open-weight models control robots in simulation and the real world. Running locally on Qwen3.8-Flash-Next, it scored 31.39 on RoboDojo, edging past GPT-6 Astra's baseline at roughly 1/20th the cost.
NVIDIA just introduced Cosmos Framework and Cosmos NIM microservices to build and deploy physical AI world models for robotics, AVs, and surgical simulation. It features new Cosmos 3 Nano checkpoints capable of generating synchronized multi-camera video and LiDAR data. With speculative decoding, the NIM containers deliver 2–3x faster inference on H200 GPUs.
Meta teamed up with Chinese robotics firm Fourier to build Max, a 5-foot, 106-pound humanoid robot designed to handle household chores. Built in just three months and powered by Nvidia’s Jetson Thor chip, the prototype gives Meta a physical testbed to train real-world AI and take on Tesla’s Optimus.
Figure AI CEO Brett Adcock just revealed that its next-gen humanoid, Figure 04, has reached full design lock with parts already shipping. Backed by $2B in cash and a $3.5B compute deal, the company is burning millions weekly to feed human video data into its Helix 3.0 "baby AGI" model.
Unitree just unveiled the H2-X, a humanoid robot dubbed the Kung Fu Legend Guardian designed for service and security. The robot boasts up to 8 hours of battery life and can perform martial arts movements. Unitree also reminded users to operate the machine safely and responsibly.
Researchers just deployed PingpongDuo, a collaborative framework that lets two Unitree G1 humanoid robots play table tennis doubles. Operating in tight quarters, the pair alternatingly returned the ball, cleared hitting space for each other, and sustained a 10-shot rally. Why it matters: Teaching agile robots real-time teamwork without bumping into each other is a major step toward shared physical workspaces.
HumanX just open-sourced a full-stack framework that turns standard human videos into agile interaction skills for humanoid robots. Tested on the Unitree G1 in simulation, it reproduces basketball pump-fakes, shots, and soccer kicks from monocular footage without motion-capture suits. It basically gives developers a way to train physical robot policies directly from everyday video.
Shenchen Kaidong just built an autonomous robot practice partner called XA1 that predicts shuttle trajectories and repositions itself to return shots without any teleoperation. It handles clears, net drops, and smashes with an over 96% hit rate, even setting a Guinness World Record with 1,452 consecutive counter-hits. Why it matters: badminton is a brutal benchmark for embodied AI, proving robots can coordinate vision and fast mobility in real-time.
Runway just debuted Praxis-1, an open-weight world action model that adapts to new robots with light fine-tuning, alongside an end-to-end campaign agent called Runway Ads. Co-CEO Cristóbal Valenzuela warned that the US is still obsessed with LLMs while China pours token usage into video models, predicting live physical AI deployments are 12 to 18 months away.
Skild AI just applied AlphaGo-style self-play to humanoid robots using its S1 model. Trained inside NVIDIA's Isaac Sim with the single goal of scoring, the system simulated 140 years of matches to master dribbling and tackling before successfully transferring into a physical robot.
Black Forest Labs just released FLUX 3 Action, an open-weight 7B world-action model designed for robotics. It achieves a 42.2% success rate on the RoboLab benchmark, beating Cosmos 3 Nano while using less than half the parameters and running up to 3.95x faster on GPUs. When paired with GPT 6 Astra in a hybrid policy, it cuts cost per successful task by 29% by handling physical execution locally.
Figure just introduced Helix 2.5, showing its humanoid robots can walk into 30 unfamiliar homes and clean up without prior mapping or fine-tuning. Pretrained on massive human behavioral data, the model boosted zero-shot success rates from 9% to 56% across tasks like tidying toys, folding towels, and making beds. Why it matters: it proves whole-body physical intelligence can transfer from human experience rather than being trained from scratch per home.
Odyssey just introduced Odyssey-3, a foundation world model that learns physics and causal dynamics directly from visual data. Instead of building narrow task-specific systems, a single frozen model can pilot robot arms, humanoids, drones, and self-driving cars with just tens of hours of control data. Why it matters: Generalist physical AI might finally be getting its GPT moment.
UBTECH just launched production at its new super factory in Liuzhou, China, capable of building over 10,000 humanoid robots annually. The flexible line cranks out a Walker S or Cruzr bot every 10 minutes, with humanoid robots even helping assemble their own peers. It marks a clear shift from lab prototypes to true industrial scale.
FANUC just partnered with Google Cloud to unveil the AI Welding Agent, powered by Gemini Enterprise. Factory operators can simply snap a photo of engineering drawings with a tablet teach pendant, and the AI automatically generates motion code and welding parameters with zero teaching. It is scheduled to ship in late December.
Sam Altman just confirmed that OpenAI will "definitely" build humanoid robots, envisioning personal robots for everyone down the line. Job postings show the company is already assembling teams to manage data collection facilities and robot hardware. Why it matters: Moving beyond software puts OpenAI into direct hardware competition with Figure and Tesla while securing proprietary real-world training data.
Hugging Face’s Pollen Robotics just unveiled Microduck, a sub-10-inch bipedal robot available for preorder at $399. The open-source bot can rollerskate, retrieve items, and follow laser pointers, all while generating a permanent unique voice on its very first boot. It is basically an affordable, hackable robotic pet built for both play and AI experimentation.
Skild AI just launched S1, a robotics foundation model that learns unseen, 10-minute tasks from a single video prompt without any fine-tuning. Using in-context learning similar to LLMs, it matches conventional VLA models that typically require 50 to 100 hours of training data. It is currently being rolled out to select industrial partners.
Generalist AI just unveiled GEN-1.5, a robot foundation model that learns new physical tasks from a single 3-to-12-second demonstration without any gradient updates or fine-tuning. Across 10 manipulation tasks, it hit an average 59% success rate via in-context physical prompting, rising to 83% with five minutes of data. It basically brings GPT-3-style in-context learning to real-world robotics, though behaviors remain short-horizon for now.
