Starting viewport
Unitree H1 walking in your tab on its maker's published policy. MuJoCo physics, ONNX inference, no install.
WASD or drag the pad to steer · Q E turn · P push · Drag to orbit · drop a URDF or MJCF on the viewport
Two tiers, one model file
01 The tab
Free, instant
- Physics
- MuJoCo 3.14, official WebAssembly build
- Render
- three.js, geoms read from the compiled model
- Input
- MJCF from the model's repo, or your own URDF/MJCF
- Policies
- 3 published walking policies replayed with onnxruntime-web, contract copied from each maker's sim2sim script
- You can
- steer a walking policy, pose joints, push, go limp, record WebM
- Limits
- one robot, CPU, single thread; inference only, no learning
02 Cloud GPUs
Queued on request
- Physics
- MuJoCo Warp on NVIDIA GPUs
- Training
- mjlab (Isaac Lab API) for locomotion and tracking; Newton for contact-rich work
- Scale
- thousands of parallel environments
- Output
- policy checkpoint, obs/action contract, eval video
- Limits
- does not run in a browser; needs CUDA via Warp
mjlab and Newton are GPU frameworks built on NVIDIA Warp. They do not run in a tab and this site does not pretend they do. What connects the tiers is the file: both load the same MJCF, so a policy trained on GPUs replays against the same model you poke here. Build a training job.
Instruments
Flagship open platform · Hyperspawn's own robot · featured, not ranked

Hyperspawn · Hyperspawn
● hyperspawn/dropbearHyperspawn Robotics
Train Dropbear.
Full-size open-source humanoid from Hyperspawn and Pointblank: FDM-printed structure on 20x20 aluminium extrusion, MyActuator CAN actuators, an ESP32 node per limb and a Jetson Orin brain. CAD, BOM, printables, firmware, URDF/MJCF/USD and simulation walking policies are public; a pre-assembled kit is on preorder.
- Sim model
- Hyperspawn/dropbear_mjcf · 180 MB
- Published policies
- 2 in the zoo
- Stand-hold bench
- fell · 2026-09-24
- DOF
- 32
Robots with an open train verb · 48
Bring your own →| Robot | Maker | Browser model | In the tab | Download | Train verb |
|---|---|---|---|---|---|
| PNDbotics | google-deepmind/mujoco_menagerie | PD hold | 164 MB | live | |
| Chestnut Robotics (formerly TetherIA) | not packaged yet | - | - | live | |
| AgiBot Lingxi X1agibot/x1 | AgiBot (Zhiyuan Robotics) | not packaged yet | - | - | live |
| Stanford (Tony Zhao et al.) | not packaged yet | - | - | live | |
| Pollen Robotics | not packaged yet | - | - | live | |
| Apptronik | google-deepmind/mujoco_menagerie | PD hold | 68 MB | live | |
| Annin Robotics | not packaged yet | - | - | live | |
| Menlo Research (Asimov) | not packaged yet | - | - | live | |
| UC Berkeley Hybrid Robotics Lab | google-deepmind/mujoco_menagerie | PD hold | 26 MB | live | |
| UC Berkeley Hybrid Robotics Lab | not packaged yet | - | - | live | |
| Bittle / Bittle Xpetoi/bittle | Petoi | not packaged yet | - | - | live |
| Booster Robotics | not packaged yet | - | - | live | |
| Booster Robotics | google-deepmind/mujoco_menagerie | walks · policy | 8.1 MB | live | |
| Agility Robotics | google-deepmind/mujoco_menagerie | PD hold | 4.9 MB | live | |
| Hyperspawn Robotics | Hyperspawn/dropbear_mjcf | PD hold | 180 MB | live | |
| Duke University General Robotics Lab | not packaged yet | - | - | live | |
| EngineAI PM01engineai/pm01 | EngineAI (Zhongqing Robotics) | not packaged yet | - | - | live |
| Fourier (Fourier Intelligence) | not packaged yet | - | - | live | |
| Fourier (Fourier Intelligence) | not packaged yet | - | - | live | |
| Fourier (Fourier Intelligence) | google-deepmind/mujoco_menagerie | PD hold | 61 MB | live | |
| The Robot Studio & Hugging Face | not packaged yet | - | - | live | |
| NASA Jet Propulsion Laboratory | not packaged yet | - | - | partial | |
| K-Scale Labs | not packaged yet | - | - | live | |
| Kangaroopal/kangaroo | PAL Robotics | not packaged yet | - | - | live |
| Alexander Koch / Hugging Face LeRobot | not packaged yet | - | - | live | |
| Carnegie Mellon University | not packaged yet | - | - | live | |
| Leju Kuavoleju/kuavo | Leju Robotics | not packaged yet | - | - | live |
| SIGRobotics UIUC / Hugging Face LeRobot | not packaged yet | - | - | live | |
| ROBOTIS | not packaged yet | - | - | live | |
| Source Robotics | not packaged yet | - | - | live | |
| Poppy Project (Inria Flowers) | not packaged yet | - | - | live | |
| Stanford Student Robotics / Hands-On Robotics | not packaged yet | - | - | live | |
| Rainbow Robotics | not packaged yet | - | - | live | |
| Pollen Robotics (Hugging Face) | not packaged yet | - | - | live | |
| ROBOTIS | google-deepmind/mujoco_menagerie | PD hold | 45 MB | live | |
| RoboParty (Shanghai RoboParty Technology) | not packaged yet | - | - | live | |
| The Robot Studio / Hugging Face | not packaged yet | - | - | live | |
| The Robot Studio / Hugging Face | not packaged yet | - | - | live | |
| Open Dynamic Robot Initiative | not packaged yet | - | - | live | |
| PAL Robotics | google-deepmind/mujoco_menagerie | PD hold | 7.6 MB | live | |
| Stanford University | google-deepmind/mujoco_menagerie | PD hold | 65 MB | live | |
| Stanford / Columbia / Toyota Research Institute | not packaged yet | - | - | live | |
| Unitree Robotics | google-deepmind/mujoco_menagerie | walks · policy | 19 MB | live | |
| Unitree Robotics | not packaged yet | - | - | live | |
| Unitree Robotics | google-deepmind/mujoco_menagerie | walks · policy | 14 MB | live | |
| Unitree Robotics | not packaged yet | - | - | live | |
| Unitree Robotics | not packaged yet | - | - | live | |
| K-Scale Labs | not packaged yet | - | - | live |