DEPLOYDatabase

Brain

RT-X / Open X-Embodiment

Google DeepMind's Open X-Embodiment initiative: the seminal cross-embodiment robot learning dataset and models (RT-1-X, RT-2-X) trained across 22 robot morphologies. Demonstrated positive transfer across robot platforms. Apache 2.0 license.

Research model · Maturity: Research · Open source

Machine-readable surfaces

Architecture

RT-1-X (35M) and RT-2-X (VLA) trained on Open X-Embodiment dataset

Key facts

Open X-Embodiment dataset (primary project page)
The Open X-Embodiment project page says the dataset pools 60 existing robot datasets from 34 labs and contains 1M+ real robot trajectories spanning 22 embodiments, from single arms to bi-manual robots and quadrupeds. The abstract says the collaboration covers 22 robots from 21 institutions and 527 skills (160,266 tasks). Treat counts as the collaboration's stated dataset scope (Open X-Embodiment)
RT-1-X and RT-2-X (primary project page)
The page defines RT-1-X as the RT-1 model trained on the robotics mixture and RT-2-X as the RT-2 vision-language model co-fine-tuned to output robot actions as language tokens. A caption calls RT-2-X (55B) one of the biggest models to date in the academic-lab demos. Actions are a 7-D gripper-frame vector (x, y, z, roll, pitch, yaw, gripper). The page says RT-X shows positive transfer across platforms. Treat transfer and the 55B figure as project-stated (Open X-Embodiment)
OXE trajectory dataset scope
The Open X-Embodiment project describes a dataset combining 60 existing robot datasets from 34 labs with more than one million real robot trajectories spanning 22 robot embodiments (Open X-Embodiment)
Google DeepMind cross robot learning
Google DeepMind explains that the Open X-Embodiment work studies learning across many robot types by pooling data from different robot embodiments and labs (Google DeepMind OXE overview)
Official dataset repository
The Google DeepMind Open X-Embodiment repository provides code and documentation for the unified robotics dataset format and access workflow behind the RT-X research line (Open X-Embodiment repository).

Developed by (1)

Mentioned on (3)

Registry pages whose text names RT-X / Open X-Embodiment.

AI systems (3)

Octo, OpenVLA, pi0 (and pi0.5)

Common questions

What is RT-X / Open X-Embodiment?
Google DeepMind's Open X-Embodiment initiative: the seminal cross-embodiment robot learning dataset and models (RT-1-X, RT-2-X) trained across 22 robot morphologies. Demonstrated positive transfer across robot platforms. Apache 2.0 license.
Who developed RT-X / Open X-Embodiment?
RT-X / Open X-Embodiment is credited to Google DeepMind on the DEPLOY registry. Each developer attribution is verified via primary sources.
Is RT-X / Open X-Embodiment open source?
Yes. RT-X / Open X-Embodiment is recorded as open-source / open-weights on the DEPLOY registry, meaning model weights or source code are publicly available.
What type of AI is RT-X / Open X-Embodiment?
RT-X / Open X-Embodiment is a research model, built on a RT-1-X (35M) and RT-2-X (VLA) trained on Open X-Embodiment dataset architecture on the DEPLOY registry.
What is RT-X / Open X-Embodiment's maturity stage?
RT-X / Open X-Embodiment is at the research stage on the DEPLOY maturity ladder. Research stage means active development without commercial deployments on file.
Which robots run on RT-X / Open X-Embodiment?
No robot models on the DEPLOY registry are recorded as running RT-X / Open X-Embodiment. DEPLOY wires brain-to-model connections only when the wiring is verifiable from primary sources; absence may reflect pre-deployment or unverified manufacturer claims.

Sources (4)

  1. https://robotics-transformer-x.github.io/
  2. https://deepmind.google/blog/scaling-up-learning-across-many-different-robot-types/
  3. https://robotics-transformer-x.github.io/paper.pdf
  4. https://github.com/google-deepmind/open_x_embodiment
Methodology: Verified · 4 sources (no primary) · last reviewed 2026-10-10

Verification posture

Verified

Low confidence

Review state

Stable

Last reviewed 2026-10-10

Maturity + lifecycle

Maturity stage: research

Sources by quality tier

3
unclassified
Unclassified source
1
code-repository
Code repository

The framework is documented at /methodology. Corrections at /corrections. Reviewer: DEPLOY editorial team.

Methodology surface for RT-X / Open X-Embodiment.

Canonical ID fd94b0f7-d33c-4176-a45e-fa31178b5b9b