Robot AI model
RT-1
RT-1 uses image sequences and language instructions to predict discretized arm and mobile-base controls. Its paper reports 35M parameters, and the official repository publishes pretrained checkpoints.
Records last verified Oct 13, 2026 · Send a correction
Also known as Robotics Transformer 1
- Organization
- Robotics at Google
- Type
- Policy
- Released / introduced
- Dec 13, 2022
- Public weights
- Public weights
- Licence / scope
- Apache-2.0
- Parameters
- 35M
- Base model
- EfficientNet with Universal Sentence Encoder language embeddings
- Humanoid association
- None reported
- Inputs
- vision, language
- Outputs
- discrete arm and base action tokens
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.
Disclosed training data
- RT-1 robot demonstration dataset
More robot AI models
Evidence
Sources
- [s1] RT-1 project: architecture, data and experiments ↗
Robotics at Google · official · accessed Oct 13, 2026
- [s2] RT-1: Robotics Transformer for Real-World Control at Scale ↗
Robotics at Google · paper · published Dec 13, 2022 · accessed Oct 13, 2026
- [s3] Robotics Transformer official repository and pretrained checkpoints ↗
Google Research · official · accessed Oct 13, 2026
- [s4] RT-1 paper: 35M architecture and language encoder ↗
Robotics at Google · paper · published Aug 11, 2023 · accessed Oct 13, 2026