Robot AI model
RDT-1B
RDT-1B is a diffusion transformer for language-conditioned manipulation across heterogeneous robot action spaces. The paper reports 1.2B parameters and evaluates bimanual skills after fine-tuning on ALOHA demonstrations.
Records last verified Oct 13, 2026 · Send a correction
Also known as Robotics Diffusion Transformer
- Organization
- Tsinghua University (TSAIL)
- Type
- VLA
- Released / introduced
- Oct 10, 2024
- Public weights
- Public weights
- Licence / scope
- MIT
- Parameters
- 1.2B
- Base model
- SigLIP vision encoder and T5-v1.1-XXL language encoder
- Humanoid association
- None reported
- Inputs
- vision, language, proprioception, control frequency
- Outputs
- continuous action chunks
Release dates can mark an announcement or paper introduction. Public checkpoints, code licences and weight terms are distinct; follow the publisher’s links for the stated scope.
Documented embodiments
Source-associated hardware; this list does not establish plug-and-play compatibility or commercial robot capability.
- ALOHA
Bimanual real-robot experiments.
Disclosed training data
- Open X-Embodiment subset →
- RH20T →
- DROID →
- BridgeData V2 →
- RT-1 Dataset, RoboSet and other datasets in the 46-dataset mixture
- RDT bimanual fine-tuning demonstrations
More robot AI models
Evidence
Sources
- [s1] RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation ↗
Tsinghua University · paper · published Oct 10, 2024 · accessed Oct 13, 2026
- [s2] RDT-1B project: method and robot experiments ↗
RDT team · official · accessed Oct 13, 2026
- [s3] RDT-1B model card: license, encoders and modalities ↗
RDT team · official · accessed Oct 13, 2026