Galaxea DynamicsChina2026
G0Tiny
An AI model: a vision-language-action model (VLA) takes camera images and an instruction and turns them into movements for a robot. This page describes the software, not a robot.
Key figures as stated by the developer
3 of 6 stated · 3 gaps- Model type not stated
- Controls not stated
- Inputs not stated
- Access A open download
- Parameters A 250,000,000
- Base model A SmolVLM2
How to read the figures
- 14kgA30.9lbA
- The letter in the grey square after a figure points to the source the figure came from. Robots with several sources carry several letters.
- 74.5lb
- Imperial figures are our own conversion unless the manufacturer states an imperial figure itself. The source letter always belongs to the metric figure — a conversion has no source of its own.
Versions and notes in the key figures (4)
Model type
Note Galaxea's README calls the checkpoint a "lightweight pretrained model".
Access
README: "Historical checkpoints remain available in the" G0-VLA Hugging Face repository, which includes the G0Tiny checkpoints. The repository is public and not gated.
Parameters
README: "(250M, SmolVLM2 backbone)"
Base model
README: "(250M, SmolVLM2 backbone)"
Full specifications
6 of 12 fields stated| Field | Value | Source and caveats |
|---|---|---|
| Input and output | ||
| Model type | not stated | Note Galaxea's README calls the checkpoint a "lightweight pretrained model". A |
| Inputs | not stated | |
| Controls | not stated | |
| Degrees of freedom controlled | not stated | |
| Size and speed | ||
| Parameters | 250,000,000 | A README: "(250M, SmolVLM2 backbone)" |
| Control rate | 10Hz | A Model inference layer on the robot's computer, not motor control: "on-device G0Tiny inference via TensorRT at up to 10 Hz". |
| Where it runs and what it is built on | ||
| Runs where | on the robot | A README: "on-device G0Tiny inference via TensorRT at up to 10 Hz"; the checkpoint is "designed for R1 Pro Orin edge deployment". |
| Base model | SmolVLM2 | A README: "(250M, SmolVLM2 backbone)" |
| Openness | ||
| Access | open download | A README: "Historical checkpoints remain available in the" G0-VLA Hugging Face repository, which includes the G0Tiny checkpoints. The repository is public and not gated. |
| License | CC BY-NC-SA 4.0 | B Model card of the G0-VLA repository, which holds the G0Tiny checkpoints: "license: cc-by-nc-sa-4.0". The GitHub code repository: "All content committed on or after 2026-01-04 and before 2026-06-16 is licensed under the G0 PLUS Community License (Non-Commercial + Limited Patent License)." |
| The developer’s own claims | ||
| Learns new tasks | not stated | |
| Works across robots | not stated | Note README: the checkpoint was pretrained with "R1 Pro VR teleoperation data" and is "designed for R1 Pro Orin edge deployment". A Yes only if the developer itself says the model runs on different robot models, for example robots from two manufacturers or two of one manufacturer’s models. Several units of the same robot do not count. |
Notes
- Galaxea designed the G0Tiny checkpoint for R1 Pro Orin edge deployment.
- Galaxea lists G0Tiny as a lightweight pretrained model.
- Galaxea places G0Tiny in its legacy G0/G0Plus release line; its current model is G0.5.
Sources
-
A
Manufacturer product page. github.
com/ · RetrievedOpenGalaxea/ GalaxeaVLA -
B
huggingface.
co/ · RetrievedOpenGalaxea/ G0-VLA
One line per source, not per figure: a source has one retrieval date. The letter beside a figure points to the line it came from.