Robotics
Robot models, compute, and adoption from open data
35 Robotics models tracked
World Top country for installs
Orders of magnitude Compute gap vs frontier LLMs
Robotics models
35 total| Name | Organization | Year | Task | Compute | Open Source |
|---|---|---|---|---|---|
| LBM 1.0 | Toyota Research Institute | 2,025 | — | — | — |
| GR00T-N1.5 | NVIDIA | 2,025 | — | — | — |
| V-JEPA 2-AC | Meta AI | 2,025 | — | — | — |
| Cosmos-Predict1-7B-Video2World (fine-tuned on Droid) | Meta AI | 2,025 | — | — | — |
| π0.5 (pi0.5) | Physical Intelligence | 2,025 | — | — | — |
| Gemini Robotics | Google DeepMind | 2,025 | — | — | — |
| GR00T-N1 | NVIDIA | 2,025 | — | — | — |
| Helix | Figure AI | 2,025 | — | — | — |
| π0 (pi0) | Physical Intelligence | 2,024 | A generalist robot policy that enables zero-shot and fine-tuned execution of highly dexterous, multi-stage tasks—such as folding laundry, packing groceries, and assembling boxes—across diverse robot embodiments. | — | — |
| RDT-1B | Tsinghua University | 2,024 | largest diffusion-based robotic foundation model | — | — |
| ALOHA Unleashed | Google DeepMind | 2,024 | — | — | — |
| OpenVLA | Stanford University,University of California (UC) Berkeley,Toyota Research Institute,Massachusetts Institute of Technology (MIT),Physical Intelligence,Google DeepMind | 2,024 | First open-source VLA that achieves state-of-the-art performance | — | — |
| Octo-Base | University of California (UC) Berkeley,Stanford University,Carnegie Mellon University (CMU),Google DeepMind | 2,024 | — | — | — |
| RFM-1 | Covariant | 2,024 | — | — | — |
| UniPi | Massachusetts Institute of Technology (MIT),University of California (UC) Berkeley,Google DeepMind,Georgia Institute of Technology,University of Alberta | 2,023 | — | — | — |
| RT-2-X | Allen Institute for AI,Arizona State University,California Institute of Technology,Carnegie Mellon University (CMU),Columbia University,Ecole Polytechnique F´ed´erale de Lausanne (EPFL),ETH Zurich,Georgia Institute of Technology,Google DeepMind,Google Research,Imperial College London,Korea Advanced Institute of Science and Technology (KAIST),Max Planck Institute for Intelligent Systems,Meta AI,Microsoft Research,Mila - Quebec AI (originally Montreal Institute for Learning Algorithms),NVIDIA,New York University (NYU),Princeton University,RIKEN,Shanghai Jiao Tong University,Stanford University,TU Darmstadt,University of Texas at Austin,The University of Tokyo,Toyota Research Institute,Tsinghua University,University of California (UC) Berkeley,UC Davis,University of California San Diego,University of Edinburgh,University of Freiburg,University of Illinois Urbana-Champaign (UIUC),University of Michigan,University of Montreal / Université de Montréal,University of Pennsylvania,University of Southern California,University of Washington | 2,023 | assembles open source dataset repository with over a million robot trajectories from many different robot embodiments; shows it can train a good model | — | — |
| V-PTR | University of California (UC) Berkeley,Google DeepMind | 2,023 | "combines the benefits of pre-training on video data with robotic offline RL approaches that train on diverse robot data, resulting in value functions and policies for manipulation tasks that perform better, act robustly, and generalize broadly" | — | — |
| RT-2 | Google DeepMind | 2,023 | largest model yet used for direct closed-loop robotic control, SOTA on some tasks, introduced VLAs (?) | — | — |
| RoboCat | Google DeepMind | 2,023 | — | — | — |
| Diffusion Policy | Columbia University,Massachusetts Institute of Technology (MIT),Toyota Research Institute | 2,023 | Implements a new way of generating robot behavior by representing a robot's visuomotor policy as a conditional denoising diffusion process | — | — |
| PaLM-E-12b | Google Research,Google,TU Berlin | 2,023 | — | — | — |
| ROSIE | Google,Google Research | 2,023 | — | — | — |
| RT-1 | Google,Google Research | 2,022 | first successful application of the foundation model paradigm to robotics | — | — |
| PerAct | University of Washington,NVIDIA | 2,022 | Shows the usefulness of voxel data: Transformer trained on voxelized 3D observation and action space outperforms unstructured image-to-action agents and 3D ConvNet | — | — |
| Interactive Language | 2,022 | — | — | — | |
| ProgPrompt | University of Southern California,NVIDIA | 2,022 | — | — | — |
| LATTE | Microsoft,Technical University of Munich | 2,022 | — | — | — |
| PaLM-SayCan | 2,022 | combines pretrained LLMs (Say) with learned robot value functions (Can) without end to end training | — | — | |
| Inner Monologue PaLM | 2,022 | LLM receives real-time environmental information and can replan accordingly, rather than just generating a fixed plan upfront | — | — | |
| Gato | Google DeepMind | 2,022 | first time that a single neural network with the same weights could successfully perform hundreds of completely different tasks | — | — |
| CLIPort | University of Washington,NVIDIA | 2,021 | — | — | — |
| Rubik's cube ADR robot | OpenAI | 2,019 | — | — | — |
| Dex-Net 4.0 | University of California (UC) Berkeley | 2,019 | — | — | — |
| Dexterous In-Hand Manipulation | OpenAI | 2,018 | — | — | — |
| QT-Opt | Google Brain,University of California (UC) Berkeley | 2,018 | — | — | — |
Compute gap
85 total| Model | Domain | Compute (FLOP) |
|---|---|---|
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | robotics | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
| — | frontier_ai | — |
Robotics model compute remains orders of magnitude below frontier LLM training compute.
Global robot adoption
124 totalTop 10 by installations
| Country | Installations | Year |
|---|---|---|
| World | 542,000 | 2,024 |
| China | 295,000 | 2,024 |
| Japan | 45,000 | 2,024 |
| United States | 34,000 | 2,024 |
| South Korea | 31,000 | 2,024 |
| Germany | 27,000 | 2,024 |
Top 10 by density
| Country | Robots per 1000 workers | Year |
|---|---|---|
| South Korea | 122 | 2,024 |
| Singapore | 81.8 | 2,024 |
| Germany | 44.9 | 2,024 |
| Japan | 44.6 | 2,024 |
| Sweden | 37.7 | 2,024 |
| Denmark | 32.9 | 2,024 |
| Slovenia | 31.5 | 2,024 |
| United States | 30.7 | 2,024 |
| Taiwan | 30.2 | 2,024 |
| Switzerland | 29.4 | 2,024 |
Latest year in data: 2024.
Data: Epoch AI robotics dataset (CC-BY-4.0), Our World in Data / IFR robot adoption. As of 2026-08-26.